NitrateSmart reads nitrate as a record of human impact, and the reading rests on geology: decades of U.S. Geological Survey groundwater studies fixed what a natural nitrate level looks like and where human activity takes over. The lines at 1, 3, 5 and 10 are each someone’s published finding — geology for 1 and 3, health studies for 5, federal law for 10.
On this page
The scale's two most important lines come straight from the U.S. Geological Survey, which spent decades sampling the nation's groundwater to answer one question: how much nitrate is natural, and how much is us.
That work is what makes a nitrate reading legible. Nitrate turns up in water on its own — soil microbes make it, rain and decaying matter carry it down — so its presence alone proves nothing. What the USGS established is the level at which the natural explanation runs out. Below about 1 mg/L as nitrogen, nitrate is consistent with untouched background[1]. Above about 3, it generally indicates human activity[2].
These two lines answer a question of provenance: whose nitrate is this? At 1, the water leaves the range that occurs in nature and begins to carry a human fingerprint. USGS work synthesized by Dubrovsky and colleagues in Circular 1350 places that boundary near 1 mg/L as nitrogen[1]. NitrateSmart calls 1–2.99 Human impact likely.
At 3, the fingerprint is no longer in doubt. Madison and Brunett's USGS review set the convention still used today: nitrate above 3 mg/L as nitrogen generally indicates human activities[2]. “Generally indicates” is not “proves” — the number cannot name the field, feedlot or pipe — but at this level natural background no longer explains it. NitrateSmart calls 3–4.99 Human impact clear. Neither line is a health claim; both are about where the nitrate came from.
A nitrate reading tells you human activity reached the water; it does not name the activity. But across the country the answer is not an even split. The U.S. Geological Survey's national assessment found agriculture to be the dominant driver of elevated nitrate in groundwater[1] — mainly fertilizer and animal manure, nitrogen applied to land faster than crops take it up, which leaches down and, because groundwater moves slowly, keeps surfacing for decades.
That is why the map of high-nitrate systems looks like a map of farming. The systems that climb into the human-impacted bands cluster in the Corn Belt and other intensively farmed regions, and they are disproportionately small systems drawing from shallow, recently recharged groundwater — the water most exposed to what happens on the fields above it. Larger systems on surface water or deep, blended sources tend to stay low.
Agriculture is not the only human source. Septic systems and wastewater add nitrate too, and matter locally, especially away from farm country — and no single reading can prove where a system's nitrate came from. But when a public supply reports nitrate well above background, the most likely story, the one the national data keeps telling, is an agricultural one.
Nitrate is worth reading closely because it rarely travels alone. The chemicals applied to a field leach toward groundwater along the same path, so where farm nitrate has reached the water, other agricultural inputs often have too. When USGS sampled shallow groundwater beneath farmland, it measured nitrate and pesticides in the same wells — and the compounds it found most often were herbicides: atrazine and its breakdown product deethylatrazine, metolachlor, simazine, the ones sprayed on the crops above[10].
Where the nitrate comes from manure rather than fertilizer, the fellow travelers change. It then tends to run with fecal bacteria and the pathogens they signal, along with the salts, veterinary antibiotics and hormones that animal waste carries. This is why an elevated nitrate result has long been a standard reason to test a well for bacteria too.
They do not all move identically — nitrate is unusually mobile and persistent, while phosphorus and some pesticides bind to the soil and pathogens are filtered out over distance. So a high nitrate reading does not prove any particular companion is present; it raises the odds, and marks water worth a closer look. And because most of those companions are tested for rarely, if ever, while nitrate is measured almost everywhere, nitrate ends up being an accessible index of how far agriculture has reached into the water — not just a number about itself.
Ten milligrams per liter is the only nitrate number in wide circulation, and it belongs to the law. It is the enforceable federal Maximum Contaminant Level, adopted by the Public Health Service in 1962[5] and re-promulgated by EPA in 1991[6], set to prevent infant methemoglobinemia — a condition that can keep a baby's blood from carrying enough oxygen. Infants under about six months are the population of greatest concern, and pregnancy is the other circumstance the literature concentrates on[13].
As an enforcement threshold it is sound. As a way to read nitrate it is blunt: one line, near the top of the range, for a single acute condition. Almost every American system sits below it, so “under the limit” drops the whole country into one box and discards the rest of the signal. NitrateSmart reports 10 and above as Above the federal limit and treats it as context — not as the point where nitrate starts to mean something.
Between the geological lines and the legal one sits the question neither answers: what is higher nitrate, sustained for years, actually associated with? The evidence that puts a line at 5 mg/L as N comes from Iowa. Two studies, published in 2010 and 2016, followed the same cohort of older Iowa women on public water supplies, both reporting in nitrate as nitrogen.
Women whose supply ran above 5 mg/L as N for more than five years showed higher thyroid cancer, at a relative risk of 2.59[11]. Women whose supply ran above the same 5 mg/L for four years or more showed higher bladder cancer, at a hazard ratio of 1.62[12]. The threshold is the concentration; the years are how long it was drunk.
What makes 5 hold is a pattern that shows up in both. In each study the comparison by concentration alone fell short of significance, while the comparison by years spent above 5 reached it. Two different cancers, two independent analyses, the same result: 5 earns its place as a level that matters when it persists, not as a line crossed on a single sample. That is what the band claims, and it is all it claims.
One distinction the Danish work[3][4] draws sharply, and which holds here too: the risk tracked nitrate in the water, not nitrate from vegetables. They are not the same exposure — the source makes the poison[9].
It is observational evidence: it shows associations, not proof of cause, and the difference for any one person is small. The thyroid finding in particular rests on eight exposed cases and an interval that only just excludes no effect. NitrateSmart draws its health-relevant line at 5 and calls 5–9.99 the Health-study range — the band a simple over-or-under-the-limit map makes invisible.
One threshold sits far below the rest. Temkin and colleagues proposed 0.14 mg/L as nitrogen[8] as a protective target — about seventy times lower than the legal limit — tied to a modeled one-in-a-million cancer risk and adverse birth outcomes. It marks the level below which even the most cautious risk models stop worrying.
NitrateSmart does not draw a band there. 0.14 sits inside natural background — at that level a human signal cannot be separated from nitrate that was always going to be present. A reading near it is not a finding; it is what untouched water looks like.
NitrateSmart classifies 38,762 community water systems in 47 states, serving 296.9 million people.
Share of the 38,762 systems, by band. Most sit in natural background; the colored tail is the human-impacted range a legal-limit view leaves out.
Only 182 systems reported a result at or above the federal limit. Below them, roughly eleven times as many reported between 5 and 10, and far more carried at least a human fingerprint. In all, 11,162 systems serving about 93.1 million people read at or above natural background — not a finding of harm, but a measure of how much further human activity reaches than the 182 legal exceedances alone would show.
Nitrate is reported two ways — as nitrogen or as nitrate ion — and the same water carries two different numbers depending on which is used. The federal limit is 10 mg/L as nitrogen, or about 44 as nitrate ion; to convert ion to nitrogen, divide by 4.43. Every number on this site is as nitrogen. Miss the conversion and one result can look four times worse in one source than in another.
This is not a footnote. It is why Schullehner and colleagues' 2018 study[7] — the largest study of all, at 2.7 million people — does not anchor the health line: its often-quoted “3.87” is an as-ion figure, about 0.87 as nitrogen. The study is sound; citing its number as though it were as-nitrogen is the error.
The two most common reasons are timing and averaging. A utility’s annual Consumer Confidence Report summarizes the previous calendar year’s compliance sampling — the range detected and how it sat against the limit — and it arrives the following summer. NitrateSmart carries each system’s most recent reported result within a three-year window. Where nitrate moves with season or source — common where supplies blend surface water and groundwater, or where recharge is fast — a year that averaged low can still end on a higher reading, and the two documents will not match.
Neither number is wrong; they describe different things. The report tells you how the year went. The most-recent result tells you where the water sits now — a snapshot, not a distribution, and free to move with the next sample. How the records are made comparable — non-detects, detection limits, unit conversions, multiple samples — is set out in How this is built, and the short version lives in Questions.
Read in order, each line answers a different question. Geology sets where human impact begins (1) and becomes clear (3). Health research marks a level that matters when it persists (5). The law draws its own line for its own purpose (10). And the strictest risk model (0.14) sits too low to separate signal from background at all. Each line is someone's published finding, not ours.
What the bands report is how strongly human activity, read through nitrate, has marked a water source — and, at or above 5, that the water has entered the range where health studies report associations. They are not a safety verdict, and they do not claim that any level will harm a particular person. Where an infant or a pregnancy is involved, that judgment belongs with a clinician.
Where the data comes from
Nitrate in community water systems, from the utilities’ own compliance reporting.
Results reaching back to July 2023 · current through August 9, 2026.
Educational information, not medical or personal safety advice.