Boil Water Advisories Follow a Pressure Event: The Bacteria Result Lands Days Later

A Pressure Reading Opens the Advisory, and a Laboratory Closes It

The notice that lands on your door or your phone says the water may be unsafe and that you should boil it. Read as a sentence, it sounds like a finding. Somebody tested something, and something was in it.

That is rarely how the sequence runs. In most cases the advisory is issued before any microbiological result exists, on the strength of a hydraulic reading. The bacteria test that would confirm or refute the concern has not been collected yet, and a federal rule on laboratory practice sets a floor on how long it takes to produce an answer once it is.

The advisory, in other words, is a decision made on a proxy. That is a defensible engineering choice, and it is a different thing from what the notice appears to say. The difference explains why boil water advisories start fast and end slowly.

EPA Region 8's loss-of-pressure guidance, which covers Wyoming systems and tribal systems in Utah, Wyoming and Colorado, states the trigger in one line: "Pressure loss is defined as a distribution system pressure of less than 20 pounds per square inch (psi)." The instruction that follows is not to sample and wait. It is to notify: "In order to protect your customers, immediately issue a Tier 1 Public Notice (PN) that includes a Loss of Pressure Boil Water Advisory."

Twenty psi is a structural number, not a health number. Below roughly that level a main can no longer be assumed to hold contaminants out. Groundwater around a leaking joint, soil around an excavation, whatever sits in a flooded valve box — at positive pressure these stay outside the pipe. The threshold describes the direction of flow across the pipe wall, and nothing about what is on the other side of it.

What EPA Counted for 2021

Until recently there was no federal tally of boil water advisories at all. Congress asked for one, and EPA's Office of Water published the U.S. Environmental Protection Agency report National Occurrence and Causes of Boil Water Advisories in the United States: Report to Congress in May 2024, covering calendar year 2021. It identified 4,036 advisories.

The cause breakdown is the finding that matters. Main breaks, distribution system repairs and loss-of-pressure events account for 3,221 of them. Advisories in the category EPA describes as microbial violation, contamination, sanitary risks and treatment failures account for 256. Another 454, or 11.2 percent, have no identified cause. Natural and other disasters account for 57, or 1.4 percent. Those four categories cover 3,988 of the 4,036.

The category EPA rounds to eighty percent is 79.8 percent when the two counts are divided. The category that matches what the phrase sounds like — somebody found something — is 6.3 percent.

There is a second layer of fog inside the big category. For 2,550 of those 3,221 events, EPA reports, the primacy agency did not identify whether there was pressure loss impact at all. The advisory was tied to a main break or a repair; whether the system actually dropped below the threshold was not recorded in a form the federal compilation could read.

The Chain

Here is the firing order, with the governing document at each step.

Step 1. A main fails or a crew opens a pipe. Pressure in a district falls. If it falls below 20 psi, EPA's guidance treats the system as having lost the barrier that keeps outside water outside.

Step 2. The notice goes out on a 24-hour clock, before any sample exists. A loss-of-pressure advisory is issued as a Tier 1 public notice. The Electronic Code of Federal Regulations, 40 CFR 141.202, Tier 1 Public Notice sets the deadline at "no later than 24 hours after the system learns of the violation or situation". The word situation is doing real work there. Tier 1 covers listed violations and also listed events, including, at item (7), "Occurrence of a waterborne disease outbreak, as defined in § 141.2" and other waterborne emergencies. No exceedance has to be measured for the clock to start.

Step 3. The system flushes and restores chlorine residual. EPA's guidance directs the operator to confirm that "the excess chlorine has been flushed out of the water supply" and that residuals have returned to normal before clearance samples are taken. Sampling heavily chlorinated flush water would produce a clean result that means nothing.

Step 4. Clearance samples are collected on both sides of the break. The instruction is to "Collect and submit to the lab a Total Coliform (TC) bacteriological sample from both upstream and downstream of the affected area of the distribution system."

Step 5. The laboratory clock starts, and it has a floor. The Electronic Code of Federal Regulations, 40 CFR 141.852, Analytical methods and laboratory certification fixes both the sample and the schedule. "The standard sample volume required for analysis, regardless of analytical method used, is 100 ml." And "The time from sample collection to initiation of test medium incubation may not exceed 30 hours." For the enzyme-substrate methods in common use, the rule notes that "Colisure® results may be read after an incubation time of 24 hours." Growth takes as long as growth takes. No amount of urgency compresses it.

Step 6. Clearance requires two days in a row, not one. The guidance says to "Maintain the boil water advisory until two consecutive days of “safe” TC samples have been collected, or until EPA notifies you that the boil order can be lifted." A single clean day is not enough. The second day's samples then run their own incubation.

Stack those and the structure becomes visible. The opening decision is made in minutes on a pressure reading. The closing decision waits on two rounds of bacterial growth, each with a transport allowance in front of it. The advisory is asymmetric by design: cheap to start, expensive to end.

The Clocks That Run While You Boil

Five separate federal time limits govern a pressure-loss advisory: three of 24 hours, one of 30, and the two-day clearance rule, which is 48 hours of sampling days. None of them is a judgment about how dangerous this particular break was.

The 30-hour figure is the least intuitive of the five. It does not mean a sample sits for 30 hours. It means the rule tolerates up to 30 hours between the moment water enters the bottle and the moment the medium starts incubating. In a metropolitan system with a certified laboratory on site, that gap can be under an hour. In a small rural system whose samples ride a courier to a laboratory two counties away, the allowance is what makes compliant sampling possible at all.

How Thin the Routine Measurement Is

None of the above is the routine monitoring program. Clearance samples are extra. The background measurement that is supposed to detect distribution system contamination in ordinary operation is set by the Electronic Code of Federal Regulations, 40 CFR 141.857, Routine monitoring requirements for public water systems serving more than 1,000 people, and it is scaled to population rather than to pipe.

A community water system serving 1,000 or fewer people takes one sample a month. A system serving between 96,001 and 130,000 people takes 100 samples a month. At the top of that band, that is one routine sample per 1,300 residents per month. The table's ceiling is 480 samples a month for systems serving more than 3,960,000 people — one sample per 8,250 residents at the boundary. Monitoring density falls as systems grow, which is defensible on statistical grounds and still means that the largest distribution networks in the country are characterized by a few hundred bottles of water a month drawn from a written sample siting plan.

That plan is required to identify "sampling sites and a sample collection schedule that are representative of water throughout the distribution system".

The Trigger Steps Up When the Sample Count Reaches 40

What happens after a positive is where the rule's arithmetic gets interesting. The Electronic Code of Federal Regulations, 40 CFR 141.858, Repeat monitoring and E. coli requirements provides that "The system must collect no fewer than three repeat samples for each total coliform-positive sample found" and that "The system must collect a set of repeat samples within 24 hours of being notified of the positive result." That deadline is not absolute: "The State may extend the 24-hour limit on a case-by-case basis if the system has a logistical problem in collecting the repeat samples within 24 hours that is beyond its control."

Whether that cluster of positives forces an investigation is decided by the Electronic Code of Federal Regulations, 40 CFR 141.859, Coliform treatment technique triggers and assessment requirements. The Level 1 trigger is written two ways. "(i) For systems taking 40 or more samples per month, the system exceeds 5.0% total coliform-positive samples for the month." And "(ii) For systems taking fewer than 40 samples per month, the system has two or more total coliform-positive samples in the same month."

Cross that language with the monitoring table and a step appears. A system serving 33,000 people takes 30 samples a month, so two positives trigger a Level 1 assessment. A system serving 33,001 people takes 40 samples a month, so it must exceed 5.0 percent of 40 — which is exactly 2.0 — and therefore needs three positives. One additional resident moves the threshold from two positives to three.

Run the arithmetic up the table and the effective threshold wanders. At 30 samples, two positives is 6.7 percent. At 40, three positives is 7.5 percent. At 50 samples three positives, and at 100 samples six, are both 6.0 percent. At 300 samples, 16 positives is 5.3 percent; at 480, 25 positives is 5.2 percent. The effect fades in large systems but does not reach 5.0, because the rule requires the percentage to be exceeded rather than met. The nominal number is a floor no system sits exactly on.

It is worth saying plainly what total coliform is measuring. When EPA revised this rule, the agency wrote in the Federal Register notice National Primary Drinking Water Regulations: Revisions to the Total Coliform Rule, published February 13, 2013, that "Many of the organisms detected by total coliform methods are not of fecal origin and do not have any direct public health implication." The same notice records the structural change: "Under the RTCR there is no longer a monthly maximum contaminant level (MCL) violation for multiple total coliform detections." Total coliform became a signal that something in the distribution system deserves inspection, not a contaminant with a health limit. The health limit that survives is for E. coli, and even that is defined by combinations of routine and repeat results rather than by a single reading.

What the Federal Record Does Not Hold

EPA's report is candid about its own foundations. "States are not required to report information regarding BWAs to the EPA." The agency adds that "not all primacy agencies maintain databases that track BWA occurrence and causes, and if they do, this tracking information is not always publicly available." Twenty states responded to the survey of tracking practices that informed the work.

The consequences show up inside the 2021 count. For 1,190 advisories — 29.5 percent — the record suggests the advisory was still open at the end of the calendar year. For 899, or 22.3 percent, the status could not be determined at all. A dataset in which nearly a quarter of entries have an unknown end state is not a duration dataset. It can tell you roughly how often advisories are issued. It cannot tell you how long people spent boiling.

Where This Doesn't Apply

Several things above are narrower than they may read.

The 20 psi figure comes from EPA's own loss-of-pressure guidance for systems where EPA is the primacy agency. Most public water systems are regulated by a state that has taken primacy, and those states set their own pressure standards and clearance protocols. Many use 20 psi; the number is common in state waterworks codes and in engineering practice. It is not a nationally uniform numeric standard in the way that an MCL is, so check the relevant state code rather than assume.

The two-consecutive-days clearance rule is likewise from that same EPA guidance, and it carries its own alternative in the original sentence: the advisory runs until two clear days are logged, "or until EPA notifies you that the boil order can be lifted." A primacy agency can lift earlier.

The 24-hour incubation figure applies to the specific enzyme-substrate method the regulation names in that sentence. Other approved methods in 40 CFR 141.852 — membrane filtration procedures, lactose fermentation techniques, EPA Method 1604 — carry their own incubation periods, and some are longer. The floor on turnaround is real, but the exact number depends on which method the certified laboratory runs.

The Level 1 trigger arithmetic assumes a system takes exactly its minimum required samples. Systems that collect additional samples under their siting plan change the denominator, and 40 CFR 141.853 governs when those extra samples count toward the triggers.

Finally, the 2021 figures are one year of an imperfect compilation assembled from state records of varying completeness. They are the best national picture that exists, not a census.

Reading Your Own Notice

Three questions separate the common case from the rare one, and all three are usually answerable from the notice itself.

First, does the notice name a cause? A main break, a repair or a pump station power failure puts the event in the 79.8 percent category — a precaution taken because the hydraulic barrier was interrupted. A notice that mentions a sample result, a treatment failure or a detection puts it in the 6.3 percent category, which is a different situation.

Second, does the notice say what has to happen before it lifts? Language about consecutive days of clear samples tells you the end is set by a laboratory schedule, not by repair progress. The pipe may have been fixed hours ago.

Third, does the notice tell you how to boil? The Centers for Disease Control and Prevention page How to Make Water Safe in an Emergency gives the instruction that matters: "Bring clear water to a rolling boil for 1 minute (at elevations above 6,500 feet, boil for 3 minutes)." That elevation adjustment is not decorative. Water boils cooler as pressure drops.

The failure mode worth naming is not that utilities issue advisories on pressure readings. That is the right call; waiting two days for a bacteriological answer before warning anyone would be worse. It is that the notice reads like a finding when it is a precaution, that the clock governing its end is a laboratory clock rather than a repair clock, and that the federal record is thin enough that nobody can say with confidence how many days of boiling the country did last year.

Two earlier pieces on this site follow the same gap between what a federal rule measures and what the public record holds: One Agency, Two Thresholds: $100,000 of Pipeline Damage Counts on Only One Line and Measuring an Ambulance Response From Dispatch, Not From the Moment Someone Dials.