Methodology

How we build each city report

We use public datasets from EPA and the Census Bureau. Here is which ones, how we connect them, and where they can be wrong.

Cities and utilities

We cover 16610 US cities, every town with at least 100 residents by the US Census Bureau 2024 population estimates and 2024 Gazetteer where we can place a single water system. For the 200 largest we found the community water system that serves each city in EPA's Safe Drinking Water Information System, using the cities each system reports serving, the system's name, and the population it serves, and fixed wrong matches by hand. For the rest we use EPA's community water system service area boundaries to find the system that serves the most residents, check it against EPA's records, and leave a city out when no single system serves at least half its residents or the two sources disagree. 16609 city pages meet our bar for a confident utility match and enough data to be useful; the rest stay hidden from search engines until we can verify them.

Large cities are often served by more than one utility, and suburbs by different ones entirely. When a city's main utility serves under 85% of its residents by EPA's service area boundaries, the page says so and names the others. Your water bill names your utility; if it differs from the one on your city's page, that utility's annual Consumer Confidence Report is the authoritative source for your home.

Chlorine or chloramine

We first read each utility's treatment records in EPA's database. Utilities that buy treated water from a wholesaler inherit the wholesaler's disinfectant, so we check the seller's records for those. Treatment records can lag real-world changes, so we then checked 1715 of the largest cities against the utility's own water quality report or website, and those answers override EPA's records. Each page says which kind of source it relies on.

Lead and copper

We show the most recent 90th-percentile result each utility reported under the Lead and Copper Rule: the level that 90% of sampled homes were at or below. EPA's lead action level is 15 parts per billion (ppb) today and drops to 10 ppb in November 2027 under the Lead and Copper Rule Improvements. FDA's limit for bottled water is 5 ppb. EPA's health goal for lead is zero. It describes the system, not your tap.

PFAS

PFAS results come from the EPA Fifth Unregulated Contaminant Monitoring Rule (UCMR 5), sampling 2023-2025 (results released through August 2026). We show the highest single result and the average of detections for six compounds. In April 2024 EPA set enforceable limits of 4 parts per trillion (ppt) for PFOA and PFOS, and 10 ppt for PFHxS, PFNA and HFPO-DA (GenX). In May 2025 EPA said it intends to keep the PFOA and PFOS limits, extend the compliance deadline to 2031, and rescind the other three. We compare results to the 2024 limits and label which ones may change.

Violations

We count health-based violations (maximum contaminant level and treatment technique violations) that began on or after January 1, 2016, from EPA's violation records. Monitoring and reporting violations are not included.

Hardness

Hardness data: TapWaterData Water Hardness Dataset (tapwaterdata.com), licensed CC BY 4.0. Values come from utility-reported data where available, otherwise from area estimates built on EPA and USGS Water Quality Portal samples. Outside the 200 largest cities we show hardness only when it was reported for that city's own utility, sometimes straight from the utility's annual water quality report. We convert to grains per gallon at 17.12 mg/L per grain and use the US Geological Survey hardness scale.

Radon

EPA radon zones describe counties, not houses. Zone 1 counties have a predicted average indoor level above 4 picocuries per liter (pCi/L), Zone 2 between 2 and 4, Zone 3 below 2. Homes with high levels are found in every zone, which is why EPA recommends that every home be tested, and fixed at 4 pCi/L or above. We place each city in the county at its geographic center using the US Census geocoder; cities spanning several counties can have more than one zone.

Air quality and wildfire smoke

Outdoor air comes from EPA AirData annual AQI by county, 2021-2025. For each county we average the years that have at least 300 days of AQI readings, and count bad-air days as days above 100 (unhealthy for sensitive groups or worse). Counties without an EPA monitor borrow the nearest monitored county within 80 km, and the page names that county. Smoke days come from NOAA Hazard Mapping System smoke polygons, 2021-2025: a county counts a smoke day when a medium or heavy plume covers a point inside it. Smoke overhead does not always reach the ground, so treat it as exposure risk. Indoor air is not measured city by city by any agency, so indoor advice is general guidance, not local data. Mold licensing rules were checked against each state's own licensing agency.

Recommendations and money

Product suggestions follow fixed rules from the findings (for example, chloramine leads to chloramine-rated filters, hard water to a softener or scale inhibitor), not from commissions. Some links are affiliate links, and installer quote requests may be shared with local companies who pay for introductions. See our disclosure.

Corrections

If something on your city's page is wrong, email the details and a source, and we will fix it.

Data sources