Abstract
Under almost every street in Chicago runs a short pipe that ties a building to the water main, and for most of the city no one has confirmed what that pipe is made of. This paper pairs published lead research with our own descriptive analysis of one real dataset, the City of Chicago Department of Water Management 2025 service line inventory submitted to the Illinois EPA, compiled and geocoded by Inside Climate News, Grist, and WBEZ and joined to U.S. Census ACS 2023 demographics. We describe how the replacement burden, the share of lines that are lead, galvanized requiring replacement, or unknown and suspected, is spread across 77 community areas and about 800 census tracts. Of 490,945 lines, 412,009 (83.9 percent) require replacement, and more than half have never been verified. The burden falls hardest on lower-income, majority-Black and majority-Latino neighborhoods on the South and West Sides. We report that as cross-sectional correlation, not cause. Replacement so far has reached 1.60 percent of lines and shows no pro-affluent skew. Two checks we ran did not behave as expected, a remediation-equity parallel to Washington and a 1938 redlining overlay, and we report each as it came out. The health stakes belong to the literature we cite. Our numbers describe the shape of a buried risk and the question of who lives above it.
The pipe no one looked at
Run the most ordinary errand of urban water and you cross it without knowing. Between the main down the center of a Chicago street and the tap inside a building sits a pipe, usually under an inch across. It crosses the parkway, ducks beneath the sidewalk, and enters the basement through a wall, and every gallon a household drinks, cooks with, and bathes in passes through it. Plumbers and engineers call it the service line. For most homes in Chicago, no record says what metal it is.
That blank is the subject here. A service line is hidden by design. It sits a few feet down, stays for decades, and outlasts the crews who laid it. Once a pipe is buried, its material stops being something you observe and becomes something you look up, and Chicago's records are mostly empty. The city has spent recent years trying to fill them, address by address, assembling an inventory of what sits underground beneath each building. The inventory it has so far tells a blunt story. For more than half the city's service lines, the honest answer to what is that pipe made of is that nobody yet knows.
This is not exotic plumbing. Every occupied building has at least one service line, so there are hundreds of thousands of them under the streets, roughly one per home. They went in across more than a century, laid by different crews under different building codes in different decades, and most have outlived whatever paper once recorded the material. A homeowner who wants to know what their own line is generally cannot tell by looking. The pipe enters through a basement wall and the long run of it lies under the parkway, where no one ever sees it. For the household and for the city both, the material is a fact about something you cannot inspect without digging.
The kind of story this is wants stating plainly at the outset. The risk here is literal and buried. The hazard, if it is one, is a length of metal under the street, and what we can measure is where that metal lies and who lives above it. So this is an infrastructure account before it is a health account, and the two stay separate throughout. What lead in drinking water does to a child belongs to a large body of medical research, which we report in its own voice with its own citations. Our numbers carry a narrower claim. They show how a buried risk is distributed across the map of Chicago, and they describe who carries it. That is the promise in the title, the risk under the street and who bears it, and we hold the paper to exactly that.
One dataset sits at the center of everything that follows. In 2025 the City of Chicago Department of Water Management submitted a service line inventory to the Illinois Environmental Protection Agency. Three newsrooms, Inside Climate News, Grist, and WBEZ, obtained that submission, cleaned it, geocoded it to addresses, and aggregated it to neighborhoods and census tracts before publishing the result. Their reporting found that roughly 412,000 of about 491,000 Chicago service lines require replacement, the most of any city in the United States, and that majority-Black and majority-Latino community areas carry a disproportionate share, the South Side worst of all.[10][9] We draw four pictures from the version they released. One looks at composition, what the pipes are across the whole city. The others rank neighborhoods by the share of pipe that needs replacing, then trace how that share moves with the median income of a census tract, and how it moves with a tract's racial composition. Every picture is descriptive. None is a claim about cause.
Why the dataset exists at all bears on how to read it. The full, machine-readable inventory is not something a resident can download from the city. It became part of the public record because journalists filed for it and published what they received. That cuts two ways for trust. The numbers we analyze trace back to the city's own submission to its state regulator, not to anything the newsrooms or we generated. And the data carries the shape of the city's recordkeeping, gaps included, the largest gap being that most service lines have never been checked.[9]
We raise no alarm about health in this opening, because raising it is not ours to do. The plain facts are enough to sit with. A pipe runs under nearly every Chicago street. For most of them the city does not know the material. A large and growing share, by the city's own count, will have to come out of the ground.
What this paper is and is not
Two things are stacked here. One is a synthesis of published work on lead, drinking water, and the history of lead plumbing. The other is our own descriptive analysis of a single public dataset. They carry different weight and do different work, and the prose marks which is which as it goes.
What we did not do is as important as what we did. No experiment. No scraping of our own. No interviews. We measured no water in any home, drew no samples, tested no taps. We fit no predictive model and ran no regression that adjusts one variable for another. We make no causal claim. We took a dataset that already existed, computed totals and shares, ranked neighborhoods by those shares, and described how the shares move alongside census measures of income and race. Those are descriptions of pattern, not findings about cause, and no sentence in this paper treats them as cause.
The arithmetic is modest and anyone could repeat it. We summed the material-classification columns across community areas for citywide totals. We sorted the 77 community areas by the share of lines requiring replacement to build the ranking. We computed Pearson correlations between that share and census measures at the tract level, dropped tracts into income quintiles and minority-share quartiles, and read the mean burden off each bin. Standard operations on a tidy table. None of them estimates an effect, holds a confounder constant, or projects a trend forward.
The correlation coefficient deserves a plain gloss, since several appear in the pages that follow. A Pearson r runs from -1 to +1 and measures only how tightly two quantities move together in a straight line across the units being compared, here census tracts. An r near zero means little linear co-movement. An r of -0.4 means a clear downward tendency with substantial scatter, not a clean rule. None of it establishes cause or direction of influence, and a coefficient computed across tracts says nothing certain about any single household inside a tract. When we report that burden correlates -0.413 with income, we mean tracts with higher incomes tend to carry lower burdens, with plenty of exceptions in both directions, and nothing more than that. We avoid words like effect, impact, and driver for these numbers on purpose, because those words smuggle in causation the method cannot supply.
Binning is the other operation worth a word. Sorting tracts into income quintiles or minority-share quartiles and averaging the burden within each is a way of seeing the shape of a relationship that a single correlation coefficient compresses into one number. It can reveal a gradient that is steep in one part of the range and flat in another, which a correlation alone would hide. The cost is that a bin average throws away the variation inside the bin. We use both the coefficient and the bins precisely because each catches what the other misses, and we report the bin sizes so that no average rests on a suspiciously thin group.
The dataset itself wants one careful definition, given here and referred back to later. It is the City of Chicago Department of Water Management 2025 service line inventory, specifically the April 14, 2025 submission to the state, which holds 491,705 records.[9] Inside Climate News, Grist, and WBEZ compiled that submission, cleaned it, geocoded the addresses, aggregated the result to Chicago's 77 community areas and its census tracts, and joined U.S. Census American Community Survey 2023 five-year demographics to those geographies. Our analysis works from that aggregated, geocoded version.[1] The small distance between the 491,705 records in the raw submission and the 490,945 lines that sum out of the community-area file reflects the cleaning and matching the newsrooms did before publishing, and we report totals as they appear in the file we actually analyzed.
How the data became public sets how much the numbers can bear. The city does not publish the row-level inventory. It offers the public a per-address lookup tool and a summary of tap-water results, not a downloadable table anyone can analyze. The complete, machine-readable inventory reached the public domain only because the three newsrooms requested it through public-records channels and posted the cleaned, geocoded version.[9] When we describe how lead-bearing pipe is distributed across Chicago, then, we are describing the city's own submission to its regulator, surfaced and organized by journalists, not a figure the city itself placed on a public dashboard.
One term recurs and deserves its definition up front. A place's replacement burden is the share of its service lines that the inventory classifies as needing replacement. That category sums three groups. Lines confirmed to be lead. Lines that are galvanized and require replacement. And lines whose material is unknown or merely suspected to be lead. Replacement burden, share requiring replacement, and lines that need replacing all mean that same three-part sum. The third group, the unverified lines, is the largest, which makes the burden figure partly a measure of what the city has not yet checked. The point comes back repeatedly, because it changes how every neighborhood number should be read.
Galvanized pipe falls into the replacement category even though galvanized iron is not lead. The reason is short. When galvanized iron sits downstream of a lead component, it can take up lead over time and release it back into the water later, so the inventory flags galvanized lines that were ever tied to lead for replacement instead of leaving them in place. We follow the inventory's classification and do not invent one of our own.
Two of our analyses came out against the hypothesis, and we say so before showing the cuts that worked. We meant to ask whether replacement so far has favored higher-income or lower-minority neighborhoods, the same equity-of-remediation question a published study put to Washington, DC.[7] It has not. We meant to lay the 1938 federal redlining map over today's lead burden, expecting the downgraded zones to carry more lead. They do not. Instead of quietly dropping the two checks that came back null or reversed, we give each its own section and report what we found. A paper that shows only the analyses that worked is not the honest version, and we would rather print the reversals.
Finally the boundary on health. Every statement about what lead exposure does to a child, about how much lead Chicago's children take in, or about how standard sampling can understate what a service line releases, comes from the published literature and carries a citation to it. We assert no health magnitude from the inventory, because the inventory contains no health measurement. It tells us where lead-bearing pipe sits. It does not tell us what is in anyone's glass. The patterns we report are associations within a single cross-section of data, and association is not cause.
Why a buried pipe matters, in the words of the research
The inventory describes where lead-bearing pipe lies, and that description only carries weight because of what is already known about lead. That knowledge is not ours. It comes from decades of toxicology, epidemiology, field measurement, and history, and the honest way to carry it is in the voice of the studies that built it, each with its citation. We are not trying to prove any of it from our own data, which cannot. We report what the research found and let it set the stakes for the distribution the data describe.
Start with the most consequential finding in the toxicology. There is no level of childhood lead exposure the research can call safe. In a pooled analysis of seven cohort studies covering 1,333 children, Lanphear and colleagues found blood lead associated with measurable deficits in children's intellectual function, and the association held even among children whose blood lead stayed below 7.5 micrograms per deciliter.[3] The deficit did not switch off at low concentrations. The per-unit harm, if anything, ran steeper at the bottom of the range than at the top. That result, drawn from cohorts on several continents, is the empirical basis for the no-safe-threshold conclusion, and it is why public-health agencies stopped calling any blood-lead level acceptable.
The pooled analysis did not appear from nowhere. Fifteen years earlier Needleman and Gatsonis assembled twelve modern studies of low-level lead exposure and children's IQ into a meta-analysis and found a consistent negative association across them, the joint significance for the blood-lead studies well past conventional thresholds.[4] Different designs, different cities, different measurement methods, and the direction of the relationship held. By the time Lanphear and colleagues pooled raw data from individual cohorts, they were sharpening a finding the field had already sketched, that more lead in a child tracks with lower measured cognitive function, and that the relationship reaches down into exposures once assumed harmless. The conclusion strengthened as the evidence accumulated instead of weakening, which is why we cite both.
No safe threshold matters for an infrastructure paper because it removes the comfortable cutoff. With a safe level, a city could aim to push exposure below it and call the work finished. The research says no such level exists for children. That shifts the live question away from whether a given exposure clears a bar toward how many children are exposed at all, and how much. It is what makes the simple presence of lead-bearing pipe, as distinct from any one measured concentration, a meaningful piece of risk information.
The shape of the dose-response curve deepens the point. In the pooled cohorts, the loss of intellectual function per unit of blood lead was steepest at the lowest exposures and flattened as exposure rose, so the first few micrograms cost a child more, on average, than later ones.[3] A relationship like that defeats the intuition that only heavily exposed children are at risk. It means a population spread thinly across many low-level exposures can carry a large summed burden, because the per-unit harm is front-loaded into the low range where most exposed children actually sit. For a city, that turns the count of children with any lead-bearing line in their water into a more relevant figure than the count of children above some action level, since there is no level below which the curve goes flat. The toxicology, in other words, is what makes a pipe inventory a public-health document and not merely a plumbing record.
Carry that toxicology to Chicago and the stakes turn specific. Huynh, Chin, and Kiang estimated childhood lead exposure from drinking water across the city and concluded that about 68 percent of Chicago children under six, on the order of 129,000 children, are exposed to lead in their tap water.[2] Their analysis went past the citywide figure. Majority-Black and majority-Hispanic city blocks faced higher odds of exposure while showing lower rates of blood-lead testing, and the exposure clustered along the same South Side and West Side geography that segregation drew decades earlier.[2] Two points carry into this paper. The exposure is widespread among the city's youngest residents. And it is unevenly distributed in a way that maps onto Chicago's racial geography. The scale figure is why service-line material is worth tracking at all. The distributional figure is the pattern our own descriptive analysis runs into independently when it ranks neighborhoods by burden, though we will present ours as a separate cross-sectional description, not a confirmation of their mechanism.
The testing gap in that study compounds the rest. The same majority-Black and majority-Hispanic blocks that faced higher exposure odds also showed lower rates of the blood-lead testing that would catch exposure in the first place.[2] Higher risk paired with thinner surveillance is a hard combination. The children most likely to be drinking lead-bearing water are among the least likely to be screened for it.
That mismatch deserves a moment, because it changes what kind of evidence is even available. Blood-lead screening is the standard way a city learns which children are exposed, and it is the trigger for intervention. But screening only finds what it looks at, and if it looks least where exposure is highest, the official record of childhood lead exposure will systematically understate the problem in exactly the neighborhoods that carry the most of it.[2] A child not tested is a child not counted, and the untested children are concentrated where the pipe is worst. So the surveillance gap and the infrastructure gap point the same direction, toward an undercount in the South and West Side neighborhoods this paper later finds carrying the heaviest replacement burden. That raises the value of knowing where the pipes are. When the screening of children skews away from the highest-risk areas, a physical inventory of which homes sit above lead-bearing line becomes one of the few sources of risk information that does not depend on whether a particular child happened to be tested. The pipe is in the ground whether or not anyone drew that child's blood, and an inventory counts it either way.
Why does the pipe itself matter so much, rather than whatever treatment the water gets at the plant? Because the service line is the last stretch of metal the water touches before the tap, and a lead line can add lead to the water after every other safeguard has done its job. Del Toral, Porter, and Schock studied this for the EPA's Region 5 in a field study of 32 Chicago homes. Lead service lines released lead at levels up to four times higher than standard first-draw sampling under the Lead and Copper Rule was detecting, meaning the routine compliance test could badly understate what a household with a lead line was actually getting.[5] They found something else with sharp consequences. Disturbing a line, through nearby street work or a partial replacement, raised lead in the water for an extended period afterward, on the order of years rather than days.[5] That measurement gap is exactly why an inventory of pipe material is worth assembling. If the standard tap test can miss much of what a lead line contributes, then knowing which homes have a lead-bearing line is its own form of risk information, independent of any single water sample. The disturbance finding also cuts against the easy assumption that digging pipes up is automatically and immediately protective. Replacement is itself a moment of elevated exposure that has to be managed.
That disturbance result bears directly on how a replacement program should be judged. If touching a lead line raises lead in the water for years, a partial replacement that swaps the public side and leaves the private side in the ground can leave a household worse off in the near term than doing nothing, since the water now runs through a freshly disturbed lead line into the home.[5] So the equity question this paper takes up later widens. It is whether replacement reaches every neighborhood, and whether it reaches each home all the way. A program that does the easy public half and stops at the property line scatters precisely the partial work the field study warns about, most often where households cannot afford to finish the private side themselves.
The field study also reframes Chicago's verification problem, the 51 percent of lines whose material the city has never confirmed. Del Toral and colleagues showed that a single first-draw sample, the kind the compliance rule relies on, can pass while the line still releases far more lead in water that has sat longer against the pipe.[5] If a brief test can miss the lead in a line known to be lead, then the absence of a confirmed-lead label on an unverified line tells a household very little about its real risk. The conservative classification the inventory applies, treating unknown pipe as needing replacement until proven otherwise, is the cautious reading the measurement science supports. A line the city has not checked is not a line the city can vouch for, and the cheapest test available would not settle the question even if it were run.
If the hazard was so well established, why is so much lead-bearing pipe still in the ground, and why does it cluster in older housing? The history answers that, and Rabin documents it. Lead's dangers in plumbing were recognized by the late 1800s, with physicians and some water authorities warning against lead pipe well before 1900.[8] Yet the Lead Industries Association ran a sustained, decades-long campaign promoting lead for residential plumbing to cities, plumbers, and code officials across the country, working to keep lead written into the specifications that governed new construction.[8] The campaign did not have to persuade the public. It had to persuade the institutions that wrote plumbing codes and trained the trades, and on that narrower front it largely succeeded, which is how a known poison stayed in specification long after the risk was on record. The result was that the very places building the most new housing in the middle decades of the twentieth century were, by code, installing lead.
Chicago is central to that story, and it held on longer than most. The city required lead service lines by ordinance into the 1980s, long after the hazard was documented and long after many other places had moved off the material.[8] That single regulatory fact explains why the burden is built into the physical fabric of the housing stock and not scattered at random. Where and when a neighborhood's pipes went into the ground, under which code, largely fixes what those pipes are today. A house built in Chicago during the long mandate got a lead service line not by accident or by a builder's thrift but by law. The bungalow belts and other areas built out during that stretch carry the mandate in their plumbing now, house after house, because the rule applied to all of them at once. The decisive date is a plumbing-code date, not a lending-map date, and the two fall in different decades. That gap turns out to matter for our own data, where a pattern emerges that a simpler redlining story does not predict.
Hold onto the timing. A material requirement that ran into the 1980s does not map onto the oldest housing in the city. It maps onto the housing built or replumbed across the long middle of the twentieth century, the outer residential belts that filled in well after the prewar core. That is a different geography from the one a 1938 redlining map would predict. The pipe follows the plumbing code and the building boom, not the color of a Depression-era lending map.
Two more studies set method, not stakes, and they matter because our analysis borrows their design. Nigra and colleagues ran a geospatial assessment of lead service lines across 2,083 census tracts in New York City, joining the tract-level presence of lead or unknown lines to racial and ethnic composition and to social-vulnerability indicators.[6] Tracts with higher Hispanic and Latino population shares were about 15 percent more likely to carry lead or unknown service lines, and lead-line risk tracked social vulnerability more broadly.[6] That is almost exactly the design our Chicago data supports, a tract-level association between the share of lead-bearing or unverified line and the demographics of the tract, reported as correspondence rather than cause. When we describe Chicago's income and racial gradients later, we are running the New York design on Chicago's inventory and presenting our correlations the way that study presented its own.
The New York study shapes our reading twice over. It treated lead and unknown lines together instead of counting only confirmed lead, on the grounds that an unverified line is itself a category of risk a city has not ruled out, and our replacement-burden measure makes the same choice for the same reason. It also reported its racial and ethnic findings as tract-level associations without claiming that demographic composition produced the distribution of pipe, which is the discipline we keep. A published analogue matters here. It means the design we apply to Chicago is not something improvised to fit a conclusion but a method already used and peer-reviewed for a comparable city.
Borrowing the design also borrows its limits, and it is fair to name them as ours. A tract-level correlation between pipe and demographics is an ecological association. It describes tracts, not households, and it cannot say that any individual minority resident is more likely to have a lead line than any individual white resident in the same city, only that tracts with more minority residents tend to carry more lead-bearing or unverified line. The New York analysis lived with that limit because the tract is the unit at which a city plans and a regulator reports, and so do we. The point of running an established design instead of a novel one is precisely that its strengths and its weaknesses are already known. We inherit a method whose ceiling is descriptive geography, which is exactly the ceiling this paper claims to work under.
The other methodological anchor concerns what happens after a city decides to replace its pipes, and it sets up one of our reversals. Baehler and colleagues examined more than 3,400 lead service line replacements in Washington, DC between 2009 and 2018, asking whether full replacement reached every neighborhood equally under the program's design.[7] It did not. The program split the cost so that homeowners paid for the private side of the line, roughly 2,000 dollars out of pocket, and full replacement turned out less common in lower-income and higher-minority neighborhoods, where fewer households could absorb that private share.[7] It is a clean example of how a remediation program can widen a disparity rather than close it, depending entirely on who is asked to pay for the work on private property. We put that question to Chicago's own replacement numbers, and we report later that the DC pattern does not hold in Chicago's current figures. The DC study is the reason we knew to ask.
The DC case is also a caution against assuming that good intentions inside a remediation program produce equitable outcomes. The District was replacing lead lines, the right goal, and still ended with a pattern in which households with the most means got the most complete fixes, through cost structure and not malice.[7] That kind of design choice can sit invisibly inside a well-meaning program and steer its benefits up the income ladder, which is why the equity-of-replacement question is worth asking of any city's numbers instead of assumed to settle itself.
That is the literature. Childhood lead has no safe level. A large share of Chicago's youngest children drink water that has touched lead-bearing infrastructure. Standard sampling can understate what a lead line releases. The burden is built into older housing by a history of mandated lead plumbing. And both the geography of lead lines and the equity of their replacement are measurable, and have been measured elsewhere. Our own contribution is narrower than any of these, and it should be plain. We add nothing to the toxicology or the exposure science. We describe, for one city in one year, where the buried risk sits and who lives above it. The studies in this section are why that description is worth doing. They are not evidence for any causal claim of our own, and we do not use them as such.
What is actually under the streets
The simplest question the inventory answers is what Chicago's service lines are made of, taken all together. Summed across the 77 community areas, the file holds 490,945 service lines.[1] Of those, 412,009 require replacement, which is 83.9 percent of every service line in the city.[1] More than four in five lines, by the city's own classification, are slated to come out of the ground. That is the headline, and it matches the figure the newsrooms reported from the same inventory, roughly 412,000 of about 491,000 lines needing replacement, the most of any city in the country.[10]
Under that 83.9 percent is where the texture lives. Confirmed lead lines are 30.7 percent of the total, or 150,856 lines.[1] Galvanized lines requiring replacement, a much smaller group, are 2.1 percent, or 10,525 lines.[1] Lines confirmed not to be lead are 16.1 percent, or 78,936 lines.[1] And the largest single category, larger than any of the others, is unknown or suspected lead at 51.1 percent, or 250,628 lines.[1]
That last figure is the one to stop on. The biggest category is not confirmed lead. It is unverified pipe. For 250,628 of Chicago's service lines, more than half of them, the city has not physically established what material is in the ground. Those lines land in the replacement burden because the inventory treats unknown material conservatively, classifying it as needing replacement until proven otherwise, not because each one has been confirmed lead. This colors how every later number reads. When a neighborhood shows a 95 percent replacement burden, that does not mean 95 percent of its lines are confirmed lead. It means 95 percent are lead, galvanized requiring replacement, or simply unverified. The phrase requires replacement is conservative by policy, a precautionary label on pipe of unknown material, not a tally of confirmed lead. Put plainly, Chicago does not know what most of its service lines are made of, and the inventory is in large part a map of that not-knowing.
One internal check passed cleanly. The inventory carries a precomputed requires_replacement count for each community area. Summed across all 77 areas, that column came to 412,009, the same number we get by adding confirmed lead, galvanized requiring replacement, and unknown or suspected across the city.[1] The two routes to the total agree to the line. That does not validate the underlying field classifications, which are the city's. It confirms that the replacement-burden figure is internally consistent with its three components in the file we analyzed.
The size of the unverified category changes what kind of document this inventory is. Read one way, a service-line inventory looks like a finished map of where the lead is. This one reads better as a record of how much checking remains. Confirmed lead and confirmed not-lead together account for fewer than half of all lines, 30.7 percent and 16.1 percent.[1] The majority sits in the unknown column, waiting on the physical verification that would move each line into one of the confirmed categories. Any neighborhood comparison built on the burden figure inherits that condition, because part of what separates a high-burden area from a low-burden one is how much of its pipe the city has actually gotten around to checking.
Most of Chicago's 490,945 service lines have never had their material verified
Composition already hints at geography, because the not-lead share is far from evenly spread. The neighborhoods with the cleanest pipes, by the share of lines confirmed not to be lead, sit downtown and on the near sides. Near South Side leads the city at 88.7 percent confirmed not-lead, and The Loop follows at 84.3 percent.[1] These are places where much of the building stock is newer, and where, accordingly, much of the service line has been confirmed as something other than lead. At the other end, the lowest not-lead shares sit in the outer bungalow belts on the South and Southwest Sides. Calumet Heights has only 3.5 percent of its lines confirmed not-lead, West Lawn 4.0 percent, Chicago Lawn 4.1 percent.[1] In those neighborhoods the share of pipe that is lead-bearing or unverified runs to the mid-nineties, roughly nineteen of every twenty lines, and the confirmed-clean share barely registers. Where the housing is the single-family brick stock built across the middle of the last century, the pipe is overwhelmingly old, and much of it has never been checked.
Lay the two ends of the not-lead ranking side by side and the split is stark. The six community areas with the highest confirmed-clean shares are Near South Side at 88.7 percent, The Loop at 84.3, Oakland at 72.7, Near North Side at 68.9, Near West Side at 60.4, and Lincoln Park at 49.8.[1] With the lone exception of Oakland, every one of them is a downtown or near-side area, and most carry high incomes, Lincoln Park at a median household income of 137,504 dollars and Near North Side at 122,602.[1] The six community areas with the lowest confirmed-clean shares are the mirror image, Gage Park at 5.2 percent not-lead, Roseland at 4.9, Ashburn at 4.6, Chicago Lawn at 4.1, West Lawn at 4.0, and Calumet Heights at 3.5.[1] All six are 90 percent minority or more, and all are South or Southwest Side.[1] At the clean end roughly half to nearly nine in ten lines are confirmed lead-free. At the heavy end roughly one in twenty or fewer are. The same two columns, read together, are the whole spatial story of the inventory in miniature.
That contrast is not yet an equity finding, because the confirmed-not-lead share blends two different things, how much genuinely lead-free pipe a place has and how much of its pipe has been checked at all. A downtown area can post a high not-lead share both because its pipes are newer and because verification has reached it. A bungalow belt can post a low not-lead share both because its pipes are older and because verification has barely begun there. The composition data cannot pull those apart on its own, one more reason we treat the geography here as a pattern to describe rather than explain.
A word on what a community area is, since the unit does work throughout the paper. Chicago's 77 community areas are stable statistical geographies first drawn by University of Chicago sociologists in the 1920s and held fixed since, which is why a researcher can line up housing, demographics, and infrastructure on the same boundaries decade after decade. They are larger than a census tract and smaller than a side of the city, and most Chicagoans know them by name, Roseland, Gage Park, Englewood, Hyde Park. Because the boundaries do not move, a community area is a clean container for a question like this one. The pipe under Calumet Heights and the people of Calumet Heights are counted on the same lines, which is what lets the inventory and the census meet.
We assert no causal link in this section between any neighborhood trait and its burden. The point is descriptive and it sets up what follows. The composition of Chicago's service lines is dominated by pipe of unknown material. The replacement-burden figure is a conservative classification, not a confirmed-lead count. And the burden is plainly unequal across the map, lightest where the building stock is newest and heaviest in the older bungalow belts. Who lives in those heaviest-burden neighborhoods is the question the ranking takes up. For now the durable facts are four shares, 30.7 percent confirmed lead, 2.1 percent galvanized requiring replacement, 51.1 percent unknown or suspected, 16.1 percent confirmed not-lead, alongside the recognition that the city's largest single class of service line is the one it has never checked.
Where the burden concentrates
Rank Chicago's community areas by the share of their service lines requiring replacement and the top of the list barely varies. The heaviest-burden neighborhoods are, almost without exception, on the South and West Sides, and almost without exception majority-minority. This describes where the buried risk concentrates. It is not a claim that any neighborhood trait produces that risk.
Share of service lines requiring replacement, highest-burden community areas
The eight community areas shown in the chart above carry some of the heaviest replacement burden in the city. Calumet Heights sits highest at 96.5 percent of its lines requiring replacement, then West Lawn at 96.0, Chicago Lawn at 95.9, Ashburn at 95.4.[1] Roseland follows at 95.1, Gage Park at 94.8, West Pullman at 94.3, and Belmont Cragin at 94.0.[1] These are not small departures from the citywide picture. The city as a whole sits at 83.9 percent, and each of these neighborhoods sits ten to thirteen points above it, with nearly every line classified as lead, galvanized requiring replacement, or unverified. The cluster is broader than these eight. Two more areas tie near the top at 94.1 percent, East Side and Beverly, with Avalon Park at 93.8 and West Englewood at 93.7 just under, so the dozen highest-burden community areas all fall in a narrow band between 93.7 and 96.5 percent.[1] The heaviest burden is not one outlier but a dozen neighborhoods carrying nearly identical, very high shares. West Englewood, lowest of that dozen, has the lowest median household income of the group at 34,375 dollars and a population that is 99.4 percent minority, while Beverly, among the highest-burden areas, has a median income of 127,056 dollars, which is the spread of incomes this cluster contains even as its burdens stay within three points of one another.[1]
The demographics behind the ranking are consistent enough to state flatly. Across the ten community areas with the highest replacement burden, the average minority population share is 89.3 percent, against 73.3 percent across all 77 community areas citywide.[1] The neighborhoods with the most pipe to replace are, on average, considerably more heavily minority than the city overall. The geography is just as concentrated. Nine of the top ten community areas by burden are on the South Side.[1] The tenth is Belmont Cragin, a West Side community area that is 79.1 percent Hispanic.[1]
The other end of the ranking sharpens it. The lightest-burden community areas cluster downtown and on the near sides. Near South Side carries the lowest burden in the city at 11.3 percent, then The Loop at 15.7, Oakland at 27.3, Near North Side at 31.1, Near West Side at 39.6, and Lincoln Park at 50.2.[1] Just above them sit Lake View at 51.8 and Douglas at 52.5.[1] With Oakland and Douglas set aside, the bottom of the burden ranking reads like a list of the city's wealthiest and most recently rebuilt neighborhoods, Lincoln Park at a median household income of 137,504 dollars, Near South Side at 124,966, Near North Side at 122,602, The Loop at 119,836, Near West Side at 110,651, Lake View at 104,135.[1] The gap from top to bottom is enormous. Calumet Heights, at 96.5 percent, sits more than 85 points above Near South Side at 11.3. Two neighborhoods in one city, and almost the entire service-line stock of the one is lead-bearing or unverified while almost none of the other's is.
It helps to picture what the heavy-burden neighborhoods physically are. Calumet Heights, West Lawn, Chicago Lawn, Ashburn, Gage Park, West Pullman. These are the brick-bungalow and brick-cottage districts that ring the South and Southwest Sides, block after block of one-and-two-story masonry houses set close together on narrow lots. Most of that housing went up across the 1920s through the postwar decades, the long middle stretch of the twentieth century when the city required lead service lines by ordinance.[8] A neighborhood that filled in during those years got lead pipe at the curb as a matter of code, house after house, and brick houses do not turn over and get rebuilt the way a downtown block does. The building form that makes these neighborhoods recognizable, the durable single-family masonry house, is the same thing that locks their original service lines in the ground for generations. That is the physical reason a ranking of replacement burden reads, from the top, like a tour of the bungalow belt.
The low-burden list is not uniformly affluent, and one entry deserves a flag. Oakland, a small South Side community area, carries a relatively low burden of 27.3 percent despite a median household income near 29,959 dollars and a population that is 94.1 percent minority.[1] It does not fit the broad pattern, where low burden tends to go with downtown geography and newer building stock. Its small size makes its share sensitive to the particulars of its housing, and it is a useful reminder that the ranking describes a strong central tendency with real exceptions at both ends, not a rule without exception.
There is a plain observation to draw, and it belongs to the equity-of-process spine of this paper rather than to any causal claim. The neighborhoods carrying the most buried risk are, by and large, the ones that have historically received the least public investment. That correspondence is descriptive. The map of heaviest replacement burden falls on the South and West Sides, the same parts of the city that decades of disinvestment and segregation shaped, and we note the overlap without asserting that one produced the other. The published literature reports the same geography from a different direction. Huynh and colleagues found estimated childhood lead exposure from drinking water clustering on the South and West Sides along historic segregation lines.[2] The newsrooms that compiled this inventory reported, from the city's own submission, that the nine community areas with the highest replacement-required share are all on the South Side.[9] Our ranking sits alongside both as an independent cross-sectional description of the same city. We offer it as correspondence, not as confirmation of a mechanism.
This is where the equity-of-process frame earns its place, so we make it explicit. A replacement burden measures risk, and it measures work that has to be done. Work raises the questions of who does it, who pays for it, and in what order. A neighborhood where 96 percent of lines need replacing faces a larger, more disruptive project than one where 11 percent do, whatever its residents can afford. When the heaviest projects fall on the neighborhoods with the least slack, the design of a replacement program decides whether it closes the gap this ranking describes or widens it. We are not grading Chicago's program here. We are establishing that the geography of the burden and the geography of disadvantage line up, so the design of any remediation effort is an equity question rather than a neutral one.
One feature of the data resists a tidy single-variable reading, and it is worth meeting head on. A couple of the highest-burden community areas are not low-income. Ashburn, fourth on the list at 95.4 percent, has a median household income of 80,909 dollars.[1] Beverly, also near the top of the burden list, has a median household income of 127,056 dollars, well above the citywide median, while still carrying a 94.1 percent burden.[1] Both are majority-minority or substantially minority, and both carry heavy burdens despite comfortable incomes. Income alone, then, is not the cleanest single predictor of where the burden falls. The burden tracks an older, more physical fact, when and under which plumbing code a neighborhood's housing went up, and it tracks the city's racial geography, two axes the tract-level cuts pull apart one at a time. For now the finding is the concentration itself. A dozen South and West Side, heavily minority community areas carry nearly all of their service lines as lead-bearing or unverified, set against near-side neighborhoods where most lines are confirmed clean.
The income gradient
Drop from the neighborhood ranking to the census tract, the finer unit, and ask how the burden tracks income. Across the 787 tracts with complete data for this comparison, the share of lines requiring replacement correlates negatively with median household income at a Pearson r of -0.413.[1] Restrict the measure to confirmed lead only, dropping the unverified and galvanized lines, and the correlation softens to -0.248, which fits, given how much of the burden is unverified pipe.[1] The sign holds in both. As tract income rises, the burden tends to fall.
Replacement burden rises as tract income falls
The gradient shows clearest when tracts are sorted into income quintiles and the mean burden read off each. The lowest-income quintile averages 80.8 percent of lines requiring replacement.[1] The second sits a touch higher at 82.6, the third at 81.9, so the bottom three quintiles bunch together in the low eighties instead of stepping down smoothly.[1] The drop comes higher up. The fourth quintile falls to 74.3, and the highest-income quintile to 58.6.[1] From the lowest-income tracts to the highest, that is a gap of 22.2 points in the mean share of pipe needing replacement.[1] The relationship is close to monotonic, with the real separation between the middle of the income distribution and the top, the wealthiest tracts standing well apart from the rest.
Put incomes on the quintiles to make them concrete. The lowest quintile has a median household income around 34,166 dollars, the highest near 134,960.[1] So the 22.2-point burden gap runs between tracts at roughly 34,000 dollars and tracts at roughly 135,000, a near-fourfold span in income matched by a fall from about four-fifths of lines needing replacement to under three-fifths.
The quintile table also shows how tightly income and race travel together in these tracts, which is worth reading off directly because it constrains everything we can claim. The lowest-income quintile averages 94.6 percent minority. The second averages 88.1, the third 75.6, the fourth 55.8, and the highest-income quintile 33.0.[1] Income falls and minority share falls in near lockstep across the same five bins. The poorest tracts in Chicago are also, on average, its most heavily minority, and the wealthiest are its least. So when the burden drops by 22.2 points from the bottom income quintile to the top, the minority share of those tracts is dropping by more than 60 points across the very same groups. The two gradients are not running on separate tracks. They are reading one population two ways, a hard limit on what either can claim rather than two independent findings.
The poverty rate, measured directly, fits the same picture but more weakly. Across the tracts, the share of lines requiring replacement correlates with the poverty rate at only +0.136, and with confirmed lead alone at +0.037, both far below the income and minority correlations.[1] That a continuous income measure correlates with burden at -0.413 while the poverty rate barely moves with it suggests the relationship is not concentrated at the bottom of the distribution, among the officially poor, but spread across the broad middle and only breaking sharply at the affluent top. It is the comfortable, mostly white, recently rebuilt tracts that stand apart on burden, not the poorest tracts that stand out as uniquely afflicted.
The shape of the gradient says as much as its size. If burden fell smoothly with every dollar of income, the five quintile means would step down evenly. They do not. The bottom three sit packed in the low eighties and the burden drops sharply only across the top two. Read descriptively, relief from buried risk concentrates at the upper end of Chicago's income distribution and does not spread across it, because being a middle-income tract rather than a low-income one buys very little reduction in burden. We read no mechanism into that shape. But a relationship that is flat across most of the range and steep only at the top is a different thing from a clean dollar-for-dollar gradient, and calling it the latter would mislead.
The plain reading stays descriptive and ties straight to the cost-and-disruption spine of the paper. Lower-income tracts sit above a higher share of pipe that needs replacing. That is the observation. Its weight for equity comes from what replacement actually involves, up-front cost, physical disruption to a property, and, where any part of the line crosses private land, the question of who pays for the private side. If replacement is paid for privately, even in part, the cost lands hardest exactly where this gradient says the most pipe sits, on the households with the least slack to absorb it. That is the structural concern the Washington study made concrete, where a private-side cost of roughly 2,000 dollars left full replacement less common in lower-income neighborhoods.[7] We raise DC here only as context for why an income gradient in buried pipe matters, not as a Chicago result. Whether Chicago's program produces a DC-style disparity is a separate question, one the replacement numbers answer with Chicago's own data.
A couple of cautions keep this in proportion. A correlation of -0.413 is a moderate association, not a tight or deterministic one. Income and burden move together across tracts in a clear direction, with plenty of scatter around the trend. Knowing a tract's income does not tell you its burden. Ashburn and Beverly are the standing reminders that comfortable income and heavy burden coexist in real neighborhoods. And income is only one axis. It is correlated with race in Chicago, as in most American cities, which means the income gradient and the racial gradient in the next section are not two independent stories but two views of an overlapping set of neighborhoods. We describe each axis on its own and do not claim to have separated them, a limit stated again when we reach it. For this section the finding is the gradient itself. The share of service lines requiring replacement falls as tract income rises, by 22.2 points from the lowest-income quintile to the highest, reported as a moderate cross-sectional association and nothing more.
The racial gradient
The same tract data carries a second gradient, and it is the stronger one. Across the 797 tracts in the racial-composition comparison, the share of lines requiring replacement correlates with minority population share at a Pearson r of +0.475.[1] Restricting to confirmed lead only, the correlation is +0.337, softer than the full-burden figure but the same sign.[1] Both run a little higher than their income counterparts, so across Chicago's tracts the burden tracks racial composition somewhat more closely than it tracks income.
Replacement burden rises with neighborhood minority share
Sorted into quartiles by minority share, the gradient steps up cleanly. The least-minority quartile averages 61.7 percent of lines requiring replacement.[1] The second rises to 70.0, the third to 81.5, the most-minority quartile to 88.3.[1] That is a climb of more than 26 points from the least- to the most-minority quartile, and unlike the income gradient it rises at every step instead of clustering and then dropping. The quartiles are close to equal in size, 200, 199, 202, and 196 tracts, so the pattern is not an artifact of a few thin bins.[1]
The inventory lets us see which part of the minority population the signal tracks, by looking at the components separately. The share of lines requiring replacement correlates with Hispanic population share at +0.373 and with non-Hispanic Black population share at +0.203.[1] Both are positive, the Hispanic association is the larger, and the overall minority correlation of +0.475 sits above either component alone. The reading we take is descriptive. The gradient rides on overall minority composition across both major groups rather than on any single group, with the Hispanic-share association the stronger single contributor in these tracts. We describe how the burden distributes across the city's racial geography. We do not attribute it to the presence of any particular population.
That the combined minority correlation runs higher than either single-group correlation matters in itself, because it means the burden does not concentrate in Black neighborhoods alone or Latino neighborhoods alone. It rises across the whole of the city's nonwhite geography. The community-area ranking showed the same thing from another angle, with majority-Black areas like Roseland and West Pullman sitting beside heavily Hispanic areas like West Lawn, Gage Park, and Belmont Cragin near the top of the burden list.[1] Two neighborhoods can differ sharply in which group predominates and still carry nearly identical burdens.
This pattern fits the plumbing-code history more than anything specific to either group. If the burden tracked one population in particular, the single-group correlation for that population would tend to dominate. Instead the broadest measure, overall minority share, carries the strongest signal, which is what one expects if the underlying driver is the age and building era of the housing and the mandate that governed it, and if Chicago's nonwhite populations were, across the middle of the century, channeled into that older mid-century stock by the same segregation the redlining maps and the private market enforced. We do not test that mechanism with these data, and we are not claiming it. We are noting that a burden which rises with minority share generally, more than with any one group, fits a story about where housing of a certain age sits and who was steered into it, better than it fits any story about a single community. The numbers describe the where. The history above is the only thing that bears on the why, and it bears on it through the housing, not through the residents.
The Hispanic-share association being the larger single component is worth one further note for honesty's sake. It does not mean Latino neighborhoods carry more lead than Black neighborhoods in any simple sense. The community-area ranking has majority-Black Calumet Heights and Roseland at the very top alongside heavily Hispanic West Lawn and Gage Park, and the citywide exposure literature locates high estimated exposure across both geographies.[2] What the component correlations capture is how each share co-varies with burden across all 800-odd tracts at once, a statistical summary that is sensitive to how each group is distributed across the income range and the housing stock. We report the components because the inventory makes them available and hiding them would be the less candid choice, but we lean on the combined minority gradient as the more stable description and decline to rank the harm by group.
The equity observation is the one the ranking surfaced, now measured at the tract level. The parts of Chicago that already carry the deepest marks of segregation and disinvestment sit above the most unreplaced lead-bearing and unverified pipe. That correspondence echoes what Huynh and colleagues reported, estimated childhood lead exposure clustering on the South and West Sides along historic segregation lines, though we present our gradient as a separate cross-sectional description, not support for their exposure mechanism.[2] It matches, in design, the New York analysis by Nigra and colleagues, who found higher-Hispanic tracts roughly 15 percent more likely to carry lead or unknown service lines and reported it as geospatial association.[6] Our Chicago gradient is the same kind of finding, a tract-level correspondence between racial composition and the share of lead-bearing or unverified line, reported as correlation and not cause.
The same limit that shadowed the income gradient shadows this one. Minority share and income are themselves correlated across Chicago's tracts, which the quintile and quartile tables make visible from both directions. The lowest-income quintile averages 94.6 percent minority. The highest-income quintile averages 33.0 percent.[1] Read the other way, the most-minority quartile and the lowest-income quintile are largely describing the same neighborhoods. So the income gradient and the racial gradient are not two independent forces we measured separately. They are two overlapping descriptions of one set of South and West Side neighborhoods. We fit no model that holds one constant while varying the other, so we make no claim about which axis matters more or whether either is independent of the rest once the others are accounted for. What we can say, and all we say, is that across Chicago's tracts the share of service lines requiring replacement rises as minority share rises, at a moderate-to-substantial correlation of +0.475, and that this gradient and the income gradient are overlapping views of one buried inequality rather than two.
Who gets fixed first, so far almost no one
Knowing where the burden sits raises the next question on its own. Where has the city actually started replacing pipe, and does that early work lean toward the neighborhoods with money? The first number to sit with is how little replacement has happened. Of 490,945 lines, just 7,874 have been replaced citywide, or 1.60 percent.[1] Roughly ninety-eight of every hundred service lines in Chicago are still in the ground exactly as the inventory found them.
We came to this section with a specific hypothesis, lifted straight from the Washington study. Baehler and colleagues found a cost-sharing replacement design skewed completed work away from lower-income, higher-minority neighborhoods.[7] We asked whether Chicago's replacements so far show the same pro-affluent or pro-white tilt. They do not.
The tract correlations run the other way. Replacement progress correlates with median household income at -0.160 and with minority share at +0.155, and with the poverty rate at +0.073.[1] All three are weak, and all three point opposite to the Washington pattern, toward marginally more progress in poorer, more heavily minority tracts. Binned by income, the lowest-income quintile has slightly more progress, 1.78 percent, than the highest-income quintile at 1.14.[1] The middle quintiles sit between, at 1.50, 1.56, and 1.54 percent, so the modest edge really does belong to the bottom of the income distribution rather than to the middle.[1]
The community areas furthest along are themselves telling. The six with the most replacement so far are West Englewood at 2.7 percent, Avalon Park at 2.4, and then Morgan Park, Austin, Auburn Gresham, and Chatham, all at 2.3.[1] Every one is a majority-Black South or West Side community area. Five of the six are more than 94 percent minority, and several sit among the lowest-income areas in the city, Austin at a median household income of 44,882 dollars, Auburn Gresham at 45,489, Chatham at 42,091.[1] These are precisely the neighborhoods a DC-style cost-sharing barrier would have left behind, and in Chicago's numbers so far they are at the front of the line. The DC cost-sharing parallel does not hold in Chicago's current figures, and we report it as not holding instead of quietly setting it aside.
The slight pro-low-income tilt earns no victory lap, and we will not spin it into a finding of fairness. Two facts hold that reading back, one plain and one technical. The plain one is scale. At 1.60 percent citywide, and with no community area above 2.7 percent, replacement has barely begun anywhere. A 0.6-point edge for the lowest-income quintile over the highest is a real direction in a program that has hardly started, not evidence of an equitable system at work. The technical one is the denominator. The 1.60 percent comes from the lines tracked for a yes-or-no replacement status, not from a guaranteed denominator of all 490,945 lines.[1] Treat it as a floor-level reading of a program in its earliest days, not a clean verdict on remediation equity. The safer summary is the plainer one. Almost the entire job is still ahead of the city, in every kind of neighborhood, and which neighborhoods get reached as the work scales up is a choice that has not yet been made at any scale worth measuring.
When the 1938 map did not explain 2025
This site is built in part on Chicago's redlining history, which made one overlay almost too tempting to skip. In the late 1930s the federal Home Owners' Loan Corporation graded neighborhoods in cities across the country by perceived lending risk, A through D, and drew the grades on color-coded maps, green for A and red for the D zones it labeled hazardous. The grades took explicit account of the race of residents, and the maps steered mortgage credit away from Black and immigrant neighborhoods for decades. A large literature ties those 1938 grades to long arcs of disinvestment that followed. Our tract file carries a Census geoid, the project already holds the 1938 HOLC graded-zone polygons for Chicago, and the two can be joined. We expected the old grades to line up with today's lead burden, the D zones carrying more suspect pipe than the A zones.
They did not. The overlay reversed the expected direction, so we dropped it from the charts and report it here as the null result it is. The join loaded 683 graded 1938 zones, 49 graded A, 160 B, 327 C, and 147 D, and matched 650 of 803 tract centroids to one of them.[1] Read by grade, the burden runs the wrong way. The D, or hazardous, zones average a 68.2 percent replacement burden, lower than the 73.3 percent in the A, or best, zones.[1] The middle grades sit highest of all, the B zones at 85.5 percent and the C zones at 81.4, so the gradient is not even monotonic, let alone aligned with the redlining frame.[1] Pooling the grades tells the same story. The combined C and D zones average 76.4 percent burden, below the 84.3 percent in pooled A and B zones, a difference of 7.9 points in the opposite direction from what the redlining frame predicts.[1] The demographic and income averages within each grade come out scrambled too. The seven A-zone tracts average a median income of 132,553 dollars and 27.7 percent minority, which fits the historical picture, but the D zones average 79,333 dollars and 74.1 percent minority while carrying the lowest burden of any grade, which does not.[1]
The mechanism the data suggest, stated descriptively and without any causal claim, is one of timing and geography. The heaviest present-day lead burden does not sit in the old urban core. It sits in the mid-century outer bungalow belts, neighborhoods like Calumet Heights, West Lawn, and Chicago Lawn, that were often graded C in 1938 or lay outside the mapped area entirely because they had barely been built. Meanwhile the older downtown and near-side tracts that drew D grades in 1938 have, over the decades since, seen much of their pipe replaced with newer, lower-lead material as the central city was rebuilt. This fits Rabin's history directly, since Chicago required lead service lines into the 1980s, well after the bungalow belts went up.[8] The pipe under a 1950s bungalow was, by city mandate, lead at installation. The pipe under a rebuilt downtown high-rise generally is not.
The method limits are real and we state them. With no geopandas available, the join used a pure-numpy centroid point-in-polygon test, assigning each tract to a 1938 grade by where its center falls, a coarse approximation for tracts that straddle a boundary. The 153 tracts whose centroids landed outside the 1938 mapped area were excluded and not imputed, the honest choice, though it leaves part of the city out of this particular cut. The lesson is the one the data force, not the one we wanted. Chicago's present-day lead burden tracks current income and racial composition far more cleanly than it tracks the 1938 map. We let the numbers overturn a tidy expectation instead of bending them to fit the redlining story, and saying so is more useful than a chart that would have implied a correspondence that is not there.
What these numbers can and cannot tell us
The limits belong in one place, as plain statements rather than scattered hedges, so a reader can see the edges of this analysis.
Everything in the four descriptive cuts is cross-sectional association from a single 2025 inventory snapshot joined to ACS 2023 five-year demographics. None of it is a causal estimate. No health outcome is measured anywhere in our numbers. We describe where the suspect pipe sits and who lives above it. We do not, and cannot from this data, show that the pipe caused anything.
The dominant material class is unverified pipe, 51.1 percent, or 250,628 lines, so requires replacement is conservative by policy and not a count of confirmed lead.[1] That single fact shadows every ranking in the paper. A tract or community area can rank high partly because it holds a lot of lead and partly because the city has not yet verified what it holds, and our data cannot fully separate the two. The burden map is, in part, a map of where verification has and has not happened.
The tract analyses drop incomplete rows rather than impute them. Of 803 tracts, 3 lack inventory percentages and 13 lack median household income, leaving 787 tracts for the income correlations and 797 for the minority-share bins.[1] The minority chart uses qcut quartiles, sized 200, 199, 202, and 196. These are small, defensible choices, named here so the bin counts are not a mystery.
The replacement-progress figure carries its own caveat, restated because it is easy to misread. The 1.60 percent comes from yes-or-no status counts tracked for only a subset of lines, not from a clean denominator of all 490,945.[1] Treat it as a floor-level signal about a program in its infancy, not a precise rate.
The collinearity limit deserves a final restatement. Income and minority share describe heavily overlapping sets of Chicago neighborhoods, and we ran no model to pull them apart. The two gradients are two views of one geography, not two independent causes.
The two analyses that reversed on us, the remediation-equity check and the 1938 HOLC overlay, are reported as null or reversed precisely because we did not bend them to the expected story. That was the point of running them in the open. A synthesis that kept only the cuts that confirmed its frame would be less trustworthy, not more.
And the boundary that governs the whole paper. The health magnitude of any of this rests entirely on the cited literature, not on our inventory. The finding that there is no safe threshold for childhood lead exposure comes from Lanphear and colleagues.[3] The estimate that Chicago's standard tap sampling understates the lead a service line releases comes from Del Toral and colleagues.[5] The estimate that a large share of the city's youngest children drink water through lead infrastructure comes from Huynh and colleagues.[2] Our numbers describe the pipe. Theirs describe why the pipe matters. The governing principle throughout has been to claim less rather than more.
Standing over the line
Strip the paper to the one thing the data durably show and it is this. The heaviest concentration of lead, galvanized, and unverified service line in Chicago sits beneath lower-income, majority-Black and majority-Latino neighborhoods on the South and West Sides. That describes where the buried risk is densest. It is not a verdict on what put it there, and we have been careful not to pretend otherwise.
The infrastructure frame and the equity-of-process spine meet at the cost of fixing the problem. Replacement carries a price, a disruption, and, for the unverified majority of lines, the prior step of even confirming what the pipe is. The neighborhoods with the most pipe to replace are the same neighborhoods with the least financial slack to absorb that work, especially where the private side of the line falls to the homeowner. So the design of any replacement program, where it starts and who pays for the private side, will decide whether the buried inequity narrows or hardens. That question is not settled. At 1.60 percent of lines replaced citywide, almost all of it is still ahead of the city, which is what makes the equity-of-process choices live rather than academic.
The health stakes we hand back to the research, where they belong. The literature is clear that there is no safe level of childhood lead exposure,[3] and that an estimated two-thirds of Chicago's youngest children drink water that has already passed through lead-bearing infrastructure.[2] Our contribution is narrower and only spatial, to show where that infrastructure is densest and who lives above it.
Which returns us to the street. Most of these pipes are invisible from the sidewalk. For more than half of them, no one has ever confirmed the material. A homeowner in West Lawn or Roseland cannot see the line under the parkway, cannot easily learn what it is, and has almost certainly not had it replaced. That is the finding. The risk is buried, it is buried unevenly, and most of it has never been checked.
[1]: Author analysis of Chicago lead water service line inventory aggregated to community areas and census tracts (with 2023 ACS demographics). Inside Climate News / Grist / WBEZ (compiled from the City of Chicago Department of Water Management 2025 service line inventory submitted to the Illinois EPA, plus U.S. Census ACS 2023 5-year). Reproducible files at rooted-forward.org/research/data/chicago-lead-service-lines-water.
[2]: Huynh BQ, Chin ET, Kiang MV. Estimated Childhood Lead Exposure From Drinking Water in Chicago. JAMA Pediatrics. 2024;178(5):473-479. doi:10.1001/jamapediatrics.2024.0133
[3]: Lanphear BP, Hornung R, Khoury J, et al. Low-Level Environmental Lead Exposure and Children's Intellectual Function: An International Pooled Analysis. Environmental Health Perspectives. 2005;113(7):894-899.
[4]: Needleman HL, Gatsonis CA. Low-level lead exposure and the IQ of children. A meta-analysis of modern studies. JAMA. 1990;263(5):673-678.
[5]: Del Toral MA, Porter A, Schock MR. Detection and Evaluation of Elevated Lead Release from Service Lines: A Field Study. Environmental Science & Technology. 2013;47(16):9300-9307. doi:10.1021/es4003636
[6]: Nigra AE, Cazabat E, Glabau D, et al. Geospatial Assessment of Racial/Ethnic Composition, Social Vulnerability, and Lead Water Service Lines in New York City. Environmental Health Perspectives. 2023;131(8):087022. doi:10.1289/EHP12276
[7]: Baehler KJ, McGraw M, Aquino MJ, Heslin R. Full Lead Service Line Replacement: A Case Study of Equity in Environmental Remediation. Sustainability. 2022;14(1):352. doi:10.3390/su14010352
[8]: Rabin R. The Lead Industry and Lead Water Pipes 'A Modest Campaign'. American Journal of Public Health. 2008;98(9):1584-1592. doi:10.2105/AJPH.2007.113555
[9]: Inside Climate News, Grist, WBEZ. How we mapped Chicago's lead pipe problem and what we learned (data and methodology). 2025. GitHub: InsideClimateNews/2025-08-chicago-lead-service-lines.
[10]: WBEZ / Chicago Sun-Times. How we mapped Chicago's lead service lines and what we learned. Aug 28, 2025.
Citations
10 sources cited.
Primary Sources
- 1.Author analysis of Chicago lead water service line inventory aggregated to community areas and census tracts (with 2023 ACS demographics). Inside Climate News / Grist / WBEZ (compiled from the City of Chicago Department of Water Management 2025 service line inventory submitted to the Illinois EPA, plus U.S. Census ACS 2023 5-year). Reproducible files at rooted-forward.org/research/data/chicago-lead-service-lines-water.
- 2.
- 3.
- 4.
Secondary Sources
- 5.
- 6.
- 7.
- 8.
- 9.
- 10.