Does Dasani have microplastics?
Yes. We bought Dasani at a grocery store in Los Angeles and it tested positive for particles consistent with microplastics, with a PET fluorescence signature.
So did the other six brands we tested. Not one came back clean. What follows is the reading, how it was produced, and — the part that matters most — what a fluorescent dye can and cannot prove.
Dasani produced the strongest signal of the seven brands. Its water is stripped by reverse osmosis before it ever reaches the bottle, which is what makes the result interesting: whatever the stain picked up, the source water is the least likely place it came from.
What the reading is — and isn’t
The number above is a fluorescence signal strength, not a particle count. It comes from staining the sample with undiluted Nile Red, pushing it through a PCTE membrane, and measuring how brightly the retained material glows across three wavelengths. A brand with a higher reading produced a stronger signal from material consistent with plastic. Nobody counted 143 of anything.
The scored brands span a narrow range, from Path at 106 to Dasani at 143. Three brands were not assigned a number in this round. We are not going to rank those three against each other to fill out a list, because we do not have the data to do it honestly.
Nile Red binds hydrophobic material. Plastic is hydrophobic — and so are lipids, oils, and some organic debris. That is why every result here is phrased as particles consistent with microplastics rather than as confirmed plastic. Definitive polymer identification requires spectroscopy, not a dye. Our full error budget, including the contamination sources we know about and the ones we suspect, is published at thewatertest.com/limitations. Read it before you treat any number on this page as gospel.
These are results from an ongoing study, not a peer-reviewed paper. Our method is adapted from published work (Meyers et al. 2022; Leonard et al. 2022); those authors have not reviewed or validated it.
We also counted particles, on camera
The reading above and the count below come from two different processes and are not two views of the same number. Counting uses diluted Nile Red on a nylon filter and asks how many particles are present. Polymer identification uses undiluted Nile Red on a PCTE filter and asks what kind of plastic they are consistent with. You cannot convert one into the other.
One 100 mL pour from a single bottle, one lot, run once. The stain indicated 10 particles, and one of them was already sitting on the filter membrane before any water touched it — visible in the video, checked before the pour. Subtract it and the honest figure is 9 particles consistent with microplastics in 100 mL.
That membrane check is not a procedural blank. A procedural blank runs clean water through the entire workflow, and we did not run one. It removes the filter’s own baseline and nothing else. The count is also a floor rather than a ceiling: a person hunting across a membrane under a digital microscope misses particles that software imaging the full filter area would catch.
It is one bottle on one day. It is not a per-liter figure for the brand, not a comparison against other brands, and not a safety finding — no regulator has set a health-based limit for microplastics in drinking water. The full write-up, including why the “1,000 particles per bottle” figure you have seen online appears in no study we can find, is here.
The other six brands
Dasani was one of seven bought together and run the same way. Reading any single brand in isolation is a mistake — the finding of the study is not that Dasani is bad, it is that the container is the common thread and every brand tested positive.
Read the full comparison, including how we identified the plastic type and why the aluminum cans scored the same as the plastic bottles, in We Tested 7 Bottled Water Brands for Microplastics.
Dasani is a registered trademark of its owner. The Water Map is not affiliated with, endorsed by, or sponsored by that company. The brand name is used here only to identify the specific product tested. We purchased the product at retail, at our own expense, and received nothing from anyone in connection with this test.
Every test we run is published, including the ones that make our own method look bad. All of our data is open — see the license.