How We Test and Source Water Data
Complete Water Guide does not run its own laboratory. Every contaminant figure published here comes from an identified third-party dataset, and each one is tied to the specific testing round that produced it. Where a brand has no published result, we say so instead of estimating.
This page explains where the numbers on this site come from, what the limits of those numbers are, and how we handle the cases where the available data disagrees with itself.
Which datasets we use
Most PFAS figures on this site trace to one of four sources. They are not interchangeable, and we name the round every time a figure appears.
| Dataset | What it measured | Scope |
|---|---|---|
| Consumer Reports, September 2020 | Total PFAS across 30 compounds | 47 bottled waters, 12 of them carbonated |
| Mamavation with Environmental Health News, September 2025 | 6:2FTS and 39 other analytes, EPA Method 1633 | 15 sparkling waters |
| FDA survey, published April 2025 | 18 PFAS compounds | 197 bottled water products, samples collected 2023 to 2024, no brands named |
| Brand and utility reports | Whatever the producer chooses to publish | Individual brands, individual water systems |
Regulatory thresholds come from the agencies that set them: the EPA’s 2024 PFAS drinking water rule for enforceable limits, and the Environmental Working Group for the 1 ppt precautionary guideline, which is an advocacy position rather than law. We label which is which.
Why we never merge the 2020 and 2025 sparkling water results
The two rounds measured different things. Consumer Reports 2020 reported total PFAS across 30 compounds. The 2025 round reported 6:2FTS, a single compound, screened out of 40 under EPA Method 1633. A brand can move between them without its water changing at all.
Only four brands appear in both, and they moved in opposite directions:
| Brand | 2020 total PFAS | 2025 6:2FTS | Direction |
|---|---|---|---|
| Topo Chico | 9.76 ppt | 1.408 ppt | down |
| La Croix | 1.16 ppt | 0.942 ppt | down |
| Perrier | 1.10 ppt | 1.064 ppt | roughly flat |
| Spindrift | 0.19 ppt | 2.621 ppt | up |
A single ranking built from both columns would be arithmetic on two different measurements. Any site that publishes one has merged datasets that cannot be merged. We publish them side by side and name the round.
How we treat brands with no published figure
Untested is recorded as untested, never as clean. This is the single most common error in bottled water coverage. Most popular brands have never been independently tested for PFAS, and an absence of results is not a result.
So a brand with no published figure appears on this site with the words “no published data” rather than a zero, a dash, or a place near the top of a ranking. If you find a site giving you a specific PFAS number for a brand that has never been tested, that number was invented.
How figures are checked before publishing
Four rules apply to every number on this site:
- Every figure is looked up, not recalled. A number that sounds obvious still gets sourced. Figures that read as verified because everything around them is verified are exactly the ones that go wrong.
- Product claims are checked against the label. Sodium, sugar and mineral figures come from the published nutrition panel or the producer’s own mineral analysis, not from another article.
- Price and cost figures are labelled as estimates where no regulator publishes a price. These are given as ranges, with the source type named.
- Where a claim cannot be sourced, it is cut rather than softened into a generalisation.
Corrections
We correct pages when the underlying data changes, and we say what changed. Two examples from this site:
- Several pages described the FDA survey as “April 2025 testing.” The samples were collected in 2023 and 2024 and only published in April 2025. Corrected across the site.
- Multiple pages stated that Waterloo and Liquid Death had no published independent data. The September 2025 round gave Waterloo 0.911 ppt and Liquid Death a non-detect. Corrected.
The EPA rule itself is also in motion: in May 2026 the agency proposed rescinding the limits for PFHxS, PFNA, GenX and the Hazard Index while keeping PFOA and PFOS at 4.0 ppt. That is a proposal, not settled law, and the 2024 limits stand in the meantime. Pages that assert the 2024 rule as permanent are out of date, including any of ours that have not yet been revised.
If you find a figure on this site you believe is wrong, tell us and include the source. Corrections are made to the page itself, not appended as a footnote.
Who writes this
Articles carry a named author and, where the subject is health-related, a named reviewer. Both link to a profile listing what that person covers. Publication and last-reviewed dates appear on every article.
- Jessica Miller, Brand Reviews Lead — bottled water brands, PFAS testing, sparkling water
- Emily Carter, Hydration & Nutrition Editor — hydration, nutrition, drinking water and health
- Michael Thompson, Water Science Writer — water chemistry, PFAS, mineral content
- David Anderson, Water Systems & Treatment Editor — treatment, municipal water quality, filtration
- Ryan Cooper, Water Quality Researcher — water quality testing, contaminants, tap water safety
How this site makes money
Some pages carry Amazon affiliate links, and we earn a commission on qualifying purchases at no extra cost to you. Affiliate relationships do not affect rankings. Brands are ordered by published test data, and several products that earn us nothing rank above products that do. Where a brand with an affiliate link has a poor test result, that result is published as it stands.
What this site cannot tell you
Three limits worth stating plainly:
- A published brand figure is one round on one production run. Bottled water is produced across multiple facilities and sources. A figure from a single test is evidence about that sample, not a permanent property of the brand.
- We cannot tell you what is in your own tap. Municipal quality varies by system and by the plumbing between the main and your glass. Your utility’s annual consumer confidence report and a certified lab test are the only ways to know.
- Nothing here is medical advice. Articles touching on infant feeding, pregnancy or a health condition say so and point you to a clinician.
Frequently asked questions
Does Complete Water Guide test water itself?
No. We do not operate a laboratory. Every contaminant figure published here comes from an identified third-party dataset, most often Consumer Reports’ 2020 testing of 47 bottled waters, the September 2025 Mamavation and Environmental Health News round under EPA Method 1633, the FDA’s survey published April 2025, or a brand’s own published water quality report.
Why do different sites give different PFAS numbers for the same brand?
Because they are usually quoting different testing rounds that measured different things. Consumer Reports 2020 reported total PFAS across 30 compounds; the 2025 round reported a single compound, 6:2FTS, out of 40 screened. A brand can appear to move sharply between them without its water changing. We name the round alongside every figure for this reason.
What does “no published data” mean on your rankings?
It means that brand has never been independently tested for PFAS in any dataset we can cite. It does not mean the water is clean, and it does not mean it is contaminated. Untested is unknown. Any article that gives a specific PFAS figure for a brand with no published testing is citing something that does not exist.
Do affiliate links affect your rankings?
No. Rankings follow published test data. Products that earn us nothing rank above products that do, and where a brand carrying an affiliate link has a poor test result, that result is published unchanged. Pages with affiliate links disclose them.
How do I report an error?
Use the contact page and include the source for the correct figure. We correct the page itself rather than appending a note, and we update the last-reviewed date when we do.
The short version
We publish other people’s lab results, we name whose results they are and when they were taken, and we do not merge datasets that measured different things. Where no test exists, we say no test exists. That last rule is the one that most separates this site from the bottled water rankings you will find elsewhere.