Evidence table · Reviewed October 6, 2026
Bottled water: measured results and missing data
Fifteen brand entries: eleven in the named 2018 cohort and four without a result in that cohort. A missing test stays missing. These are historical measurements, not today’s safety ratings.
| Brand | Separate lot means / L | Highest bottle / L | Reported brand mean / L | Evidence note |
|---|---|---|---|---|
| Nestlé Pure Life | 2,277 / 49.3 / 468 | 10,390 | 930 | Historical sampled lots; not a current product test. |
| Gerolsteiner | 1,410 / 204 | 5,106 | 807 | The 204/L lot was glass; the 1,410/L lot was plastic. |
| Aqua | 37.1 / 705 / 404 | 4,713 | Not reproduced here | Historical sampled lots; not a current product test. |
| Aquafina | 252 / 174 | 1,295 | Not reproduced here | Historical sampled lots; not a current product test. |
| Bisleri | 826 / 213 / 3.72 | 5,230 | Not reproduced here | Historical sampled lots; not a current product test. |
| Dasani | 165 / 74.6 | 335 | Not reproduced here | Historical sampled lots; not a current product test. |
| E-Pura | 686 / 20 / 37.3 | 2,267 | Not reproduced here | Historical sampled lots; not a current product test. |
| Evian | 197 / 58.2 | 256 | Not reproduced here | Historical sampled lots; not a current product test. |
| Minalba | 40.1 / 12.5 / 150 | 863 | 63.1 | Historical sampled lots; not a current product test. |
| San Pellegrino | 30.3 | 74 | 30 | Figure 1/Results reports the brand mean rounded to 30.0/L. |
| Wahaha | 156 / 66.7 / 67.1 | 731 | Not reproduced here | Historical sampled lots; not a current product test. |
| Smartwater | Not measured in this cohort | — | Not reproduced here | Not in the named 2018 cohort; no number assigned here. |
| Fiji | Not measured in this cohort | — | Not reproduced here | Not in the named 2018 cohort; no number assigned here. |
| Voss | Not measured in this cohort | — | Not reproduced here | Not in the named 2018 cohort; no number assigned here. |
| Mountain Valley | Not measured in this cohort | — | Not reproduced here | Not in the named 2018 cohort; no number assigned here. |
Method and limits
The transcription preserves the statistic reported in each field. We do not calculate a brand mean by averaging rounded lot means, assign a particle count from packaging alone, or treat an analytical count as a health-risk tier.
The 2024 Qian study examined three unnamed brands using a different method. Its average of about 240,000 micro- and nanoplastic particles/L, roughly 90% nanoplastics, is not assigned to any row above. Differences in size limits and identification prevent a simple cross-study leaderboard.
A finished-water measurement also differs from a bottle-material declaration. Glass, caps, filling processes and source water should be considered separately. The unmeasured rows here do not imply that no test can exist; they identify what this named cohort does not establish.
Original sources
- Mason, Welch and Neratko (2018): Table 3 and reported brand means
- Qian and colleagues (2024): stimulated Raman scattering study
Read the bottled-water comparison and buying decisions →
How we distinguish measurements from modeled contact assessments →