Federal drinking water law covers only systems with at least 15 service connections or 25 regularly served people, and EPA states as of September 2026 that water from private household wells is not federally regulated under it

What is happening?

The Safe Drinking Water Act builds its drinking water rules around the system, not around the person drinking. Under 42 U.S.C. 300f(4)(A) a public water system is one that provides piped water for human consumption to the public and has at least fifteen service connections or regularly serves at least twenty-five individuals, and the national primary drinking water regulations apply to public water systems. The implementing regulation at 40 CFR 141.2 adds more detail, defining the second prong as serving an average of at least twenty-five individuals daily at least 60 days out of the year. The 60-day wording belongs to the regulation, not to the statute.

A household well sits below both thresholds. The federal agencies say so on their own pages. EPA, on a page updated 2026-08-25, states that the quality and safety of drinking water from private domestic wells are not regulated by the Federal Government under the Safe Drinking Water Act, and that private well owners are responsible for delivering safe drinking water to their households. USGS states that water from domestic wells is not regulated by the federal act or, in most cases, by state laws. CDC, on a page reviewed 2026-09-01, states that the act focuses on public water systems and does not cover private drinking water sources, and that such systems therefore might not be tested regularly for contaminants.

What follows from the definition is not the absence of every federal tool. It is the absence of a standing federal framework of enforceable water quality standards, required periodic testing and required notification for the water that comes out of a household well. Testing is left to the owner, and both EPA and CDC recommend that owners test once a year for total coliform bacteria, nitrates, total dissolved solids and pH.

Whose problem is this?

The people drinking the water are the ones who have to decide whether and when to test it. The federal agencies that describe the gap are the same agencies whose statute places the well outside its standards.

Where does this problem end?

The last row is set aside on purpose. This document uses the health benchmarks that EPA and USGS already published and does not choose its own. It also does not say why the boundary was drawn where it was, because no opened source gives the reason for the fifteen and twenty-five thresholds.

The federal government does reach private wells in some ways that are outside this frame: a USDA loan fund for building and repairing household wells, the annual testing advice from EPA and CDC, and an emergency power under 42 U.S.C. 300i whose text covers underground sources of drinking water. What this document describes is the missing standing framework, not a total absence of federal involvement.

What is the state now, and what should it be?

Now

Needs a new measurementthe target state: no source opened here sets a federal numeric target for how many private wells should be tested, how often, or what share should meet health benchmarks. The one federal health objective on drinking water, Healthy People 2030 EH-03, counts only people served by community water systems and leaves private wells outside the indicator. The EPA and CDC annual testing advice is guidance with no target attached. No advocacy or professional target has been substituted in its place.

How big is it?

The number of people involved is large, but the federal sources do not agree on it and most of them do not date it. EPA says over 43 million people, around 15 percent of the population, rely on private wells, and more than 23 million households, with no reference year on the page. USGS gives the same 43 million and 15 percent, also without a year. CDC says about 1 in 8 residents and ties that to a 2018 USGS report and 2018 Census estimates, while an earlier CDC document from 2018-02-06 says about 1 in 9 without naming its base year. The one dated model, published by USGS in 2019, estimates 37.25 million people in 2000 and 37.29 million in 2010, for the contiguous 48 states only.

This document therefore uses a range, from 37.29 million to 43 million people, and does not merge the figures into one. The detail is in the population estimate attached to this page.

How often the water fails a health benchmark is measured separately. A USGS national sample of up to 2,167 domestic wells taken from 1991 through 2004 found about 23 percent with at least one contaminant above a federal drinking water standard or a USGS health benchmark. In New Jersey, where testing is required at sale or lease, 6,369 of 51,028 wells tested from September 2002 through April 2007 exceeded at least one primary standard, which is 12.5 percent of the tested wells. The two rates come from different years, methods and benchmarks and are not comparable to each other, and an exceedance is a concentration measurement, not a count of illness.

Under what conditions does it arise?

1. The law defines its unit as a system with a minimum size. A single house with its own well has one connection and serves a household, which is below both statutory thresholds, so the federal standards never attach to it in the first place.

2. Responsibility moves to the owner by default. Once the well is outside the definition, the agencies describe the owner as the party responsible for the safety of the water, and the federal instrument that remains is advice to test once a year. Whether an owner tests, and what an owner does with a bad result, is not recorded anywhere in the sources opened here.

3. States decide how far to go, and most stop at construction. In the 2016 through 2018 policy review every state had rules for how wells are built, but only about a fifth had a rule tied to property sales and only a small share covered rental housing. New Jersey is an example of a state that requires testing at sale or lease, and as of its 2008 report that law did not require treatment when a test exceeded a standard.

4. The national picture rests on old and scattered measurements. The national USGS sample of domestic well water opened here was taken between 1991 and 2004, and the population figures in current federal pages carry no year or disagree with each other, so the size of the exposure cannot be tracked over time from these sources.

What has been tried?

The tools that exist fall into three kinds: advice to owners, measurement by federal scientists, and rules set by individual states. None of the three puts a standing federal requirement on the water from a household well, and the one state requirement documented here, as of 2008, stopped at testing.

What was found?

Why is it still unsolved?

Institutional exemption — the federal act sets a floor for water systems above a size threshold, and a household well is below that threshold by construction.

The act does not say that well water is safe or unsafe. It defines which water it governs, and a household well is simply not in that set. Everything the act builds on top of the definition — maximum contaminant levels, monitoring schedules, public notice when a standard is exceeded — therefore has nothing to attach to. The federal agencies describe the result plainly on their own pages: the owner is responsible.

What replaces the federal framework is a patchwork that depends on where the well is. Every state regulates how a well is built, but construction rules speak to the hole and the casing, not to what is in the water years later. Only a minority of states tie testing to a sale, fewer to a rental, and the one state program documented in detail here required a test but, as of 2008, not treatment. A household can move across a state line and go from a required test at purchase to no rule at all, while drinking from the same aquifer.

The third part is that the gap is hard to see from above. The federal health objective on drinking water measures only community systems, so progress on that indicator says nothing about wells. The national USGS sample of well water opened here was taken more than twenty years ago, and the current federal pages give population figures that are either undated or inconsistent. A problem that is not in the national indicator and not in a recent national measurement does not show up as a trend that anyone is obliged to answer.

What observation would mean it is solved?

Candidates — (a) a standing requirement, federal or state, for periodic testing of household well water together with notification of results to the people drinking it (b) a published national measurement, repeated over time, of the share of private wells meeting health benchmarks (c) a national health objective or target that includes people on private wells.

(a) alone is weaker than it looks. New Jersey, as of its 2008 report, shows that a testing requirement can exist without any requirement to treat, and a test tied only to a sale or lease reaches a well once in many years. A requirement written into one state leaves the others where they were.

(b) alone measures without changing anything. A repeated national sample would show whether the share of wells above a benchmark is rising or falling, but a falling share could reflect where the sample was taken rather than what households drink.

(c) alone counts intentions. An objective that includes private wells would make the gap visible in the national indicator, but it would still need a measurement behind it, which brings the observation back to (b). The three have to be read together, and none of them is currently observed in the sources opened here.

What is it connected to?

These links were drawn from the scope of existing documents in this corpus. The relation is one of shared boundary, and no source opened here measures how the problems interact.

What these sources do not say

• The year behind the population figures. EPA and USGS give over 43 million and about 15 percent with no reference year, CDC gives about 1 in 8 in one place and about 1 in 9 in another, and the only dated figure covers the contiguous 48 states in 2010. The sources do not reconcile these numbers.
• How many owners actually test. No opened source gives a testing rate for private well owners. CDC says only that wells might not be tested regularly.
• Any national resampling after 2004. The opened USGS pages do not describe a later national sample of the same scale, and they do not say that none exists.
• Whether the 2,100 wells and the 2,167 wells are the same sample. The USGS summary page cites a study of 2,100 domestic wells with about one in five above a benchmark, and the 2009 report cites up to 2,167 wells with about 23 percent. The opened pages do not connect the two.
• Current New Jersey requirements. The opened New Jersey material ends at the 2008 report and does not state what the law requires today, including whether treatment is now required.
• Whether the emergency power has been used for private wells. The text of 42 U.S.C. 300i covers underground sources of drinking water but does not name private wells, and no opened source describes a case or an EPA interpretation applying it to one.
• How representative the New Jersey tests are. The state report notes errors in sampling, analysis and location records and says no state agency can confirm that every sale and lease was reported. It does not say whether wells tested at a sale resemble wells in general.
• What causes the exceedances. No opened source attributes the exceedances to natural or human sources, and this document names no pollution source.
• Private well and public supply PFAS comparison. The USGS dashboard gives the sample size but no detection rates, and no opened source compares the two.
• A federal numeric target for private wells. No opened source sets one. Healthy People 2030 EH-03 counts only community water systems.
• The position of the agencies beyond their own pages. EPA, USGS and CDC state on their own pages that the federal act does not cover private wells and that owners are responsible. That is the full published position carried here. No agency explanation of why the threshold sits where it does, and no rebuttal of these statements, was found.
• Current state policy counts. The most recent state review opened covers 2016 through 2018 and does not describe later changes.
• Health outcomes. An exceedance of a benchmark is a concentration, not a case of illness, and no opened source links the exceedance rates to disease counts.

See the evidence

The statutory and regulatory texts were read directly and carry the thresholds. The agency statements come from the public pages of EPA, USGS and CDC as opened on 2026-09-24. The measurement values come from a USGS publication page, a USGS model publication, a New Jersey state report from 2008 and an academic state policy review; state and journal sources were used for facts only. Where the federal population figures disagree, the disagreement is left visible rather than reconciled. This is a research-based definition, so observation_refs is empty and provenance_mode: press-derived.

People affected

Estimated range 37,290,000–43,000,000 As of 2010 model estimate to undated agency figure on a page updated 2026-08-25

Sensitivity The interval is the spread between two federal figures that do not share a base year or a geography, not a confidence interval. The high end is itself a floor, because the agencies say over 43 million, so the upper bound understates what those pages claim. A third federal source, CDC, gives about 1 in 8 residents tied to 2018 data in one place and about 1 in 9 with no stated base year in another; those ratios were not converted to counts here because doing so would require choosing a population denominator the sources do not supply. The count measures people drinking water outside the federal standards, not people drinking water that fails a benchmark. A USGS sample from 1991 through 2004 found about 23 percent of sampled wells with at least one contaminant above a benchmark, but that share was not multiplied into the population because the sample years, the unit of wells rather than people, and the benchmarks differ from the population figures.

Regional breakdown No source opened here gives a state-by-state count of people on private wells. The only figures available are national totals, and splitting them by state population would be proportional allocation, which is not permitted.