Bottled water: you're paying extra for the plastic

The bottled water industry sells one thing: purity. There’s a glacier on the label. There’s plastic in the water. Every lab that has gone looking with tools fine enough to see small particles has found plastic in bottled water, and part of it comes from the bottle and cap the industry charges you for.

Tap water isn’t plastic-free either. But the newest studies make the gap clearer, and it doesn’t flatter the bottle. Here’s what the research found, how much weight each number can carry, and what lowers your exposure.

Every method that looks finds plastic in bottled water

The first big survey tested 259 bottles from 11 brands, bought in 19 locations across nine countries. Mason and colleagues found “some sign of microplastic contamination” in 93% of them. After subtracting possible lab contamination, they averaged 325 particles per liter down to 6.5 micrometers, with single bottles ranging from zero to more than 10,000. The brands included Aquafina (PepsiCo), Dasani (Coca-Cola), Evian (Danone), and Nestlé Pure Life and San Pellegrino (both Nestlé). Particles under 100 micrometers were flagged with a dye, not confirmed by spectroscopy, so the authors call them “most probably plastic.” The nonprofit newsroom Orb Media bought the samples (Mason et al.).

Those counts were capped by what 2018 microscopes could see. Newer tools see far smaller particles, and the numbers jump.

• About 240,000 particles per liter. A Columbia University team used a laser imaging method, stimulated Raman scattering microscopy, on three unnamed brands bought from one large retailer. “Micro-nano plastics concentrations were estimated to be about 2.4 ± 1.3 × 10⁵ particles per liter of bottled water, about 90% of which are nanoplastics” (Qian et al.). That’s roughly 240,000 per liter, give or take 130,000.
• About 99,000 per liter, from a different lab with a different method. A 2026 study used mid-infrared photothermal microscopy, which identifies the plastic in each particle, on three major Chinese brands. It counted 9.9 ± 1.2 × 10⁴ plastic particles per liter, 64% of them nanoplastics. PET, the stuff the bottles are made of, was the dominant polymer. The team ran blanks with ultrapure lab water and reported minimal contamination (Deng et al.). Three brands is a small sample, but it lands in Columbia’s range.
• Bottled lost to tap, head to head. An Ohio State team used electron microscopy plus infrared spectroscopy, which identifies polymers down to submicron sizes, on water from four treatment plants and six bottled brands. “Bottled water had significantly higher particle concentrations of MNPs than the treated drinking water, specifically for the smaller micro- and nanoplastics” (Jamison Hart and Lenhart). The university’s summary puts it at three times as many nanoplastic particles (Ohio State). The bottled water was rich in polyamide (nylon-type plastic), PET and polyethylene.

The fight over the biggest number

The Columbia figure is contested. In a letter in the same journal, D. Materić pointed out that the study’s own supplement says its ultrapure lab water “seems to have the same level of plastic contamination compared with bottled water measured.” The team used clean, dry filters as its blank instead, a choice Materić argues makes the quantitative results “fundamentally unreliable” (Materić). The authors replied that the lab water was itself contaminated, which made it the wrong blank. They added that the three brands had different plastic profiles, which a single lab contaminant wouldn’t produce (Qian et al. reply).

Notice what that fight is about: how many particles, not whether they’re there. Since then, a separate lab using lab-water blanks got a count in the same range (Deng et al.).

The bottle and the cap are part of the source

Among the particles Mason’s team checked by spectroscopy, the most common plastic was polypropylene, at 54%. The authors note this “matches a common plastic used for the manufacture of bottle caps,” and conclude the contamination is “at least partially coming from the packaging and/or the bottling process itself” (Mason et al.).

How you treat the bottle matters too.

• Opening and closing. An Italian team opened and closed PET bottles 1, 10 and 100 times. They found “a considerable increase” in plastic particles on the bottleneck and cap surfaces, and after 100 times “the effect was impressive, especially on caps.” Squeezing the bottle, by contrast, did not significantly raise particle counts in the water. Their conclusion: the chance of swallowing plastic goes up “with frequent use of the same single-use plastic bottle, though only from the bottleneck-cap system” (Winkler et al.).
• The hot car test. A 2026 University of Nebraska study put eight leading US brands through what the authors call “an extreme yet realistic upper-bound” for a hot car: 60 °C (140 °F) with shaking, for 12 hours. That combination did the most damage. For five of the brands, nanoparticle counts rose by as much as 9.29 times. Fifteen days of daily heating, or of freeze–thaw cycles, also raised nanoparticle counts. Raman spectroscopy found PET, polyethylene and polypropylene particles, which the authors trace to “both bottle body and caps” (Hussain et al.).

One caution: the counting method sees every particle, plastic or not, and only a small subset was chemically checked, which the authors call “qualitative confirmation” (Hussain et al.). So the 9.29 means heat plus jostling shakes loose far more particles, some confirmed as bottle and cap plastic. It’s not a plastic count.

The plot twist: glass bottles with painted caps

If you’ve concluded “plastic bottle bad, glass bottle good,” France’s food safety agency, ANSES, has news. Its researchers tested water, sodas, teas, beers and wines sold in France in glass, plastic, cans, cartons and bag-in-box. “It was observed that the most contaminated containers were glass bottles” (Chaïb et al.). Glass-bottled colas, lemonades, iced teas and beers averaged around 100 particles per liter, and levels in plastic bottles and cans were five to 50 times lower. Water was the cleanest category, at 4.5 particles per liter in glass and 1.6 in plastic bottles and cartons (ANSES).

The culprit was the cap. Most particles matched the color and composition of the paint on the outside of the metal caps (Chaïb et al.). ANSES suspects the caps scratch each other in storage. In a test, water sealed with uncleaned caps had 287 particles per liter. Blowing air on the caps first cut that to 106; blowing plus rinsing got it to 87. The agency put the fix where it belongs, on manufacturers: store caps so they don’t rub, or change the paint (ANSES).

There’s pushback on the method. A comment led by Jane Muncke of the Food Packaging Forum argues that reporting particles per liter inflates results for small packages, and the glass bottles were generally smaller than the plastic ones (Muncke et al.). Fair. The study also counted microplastics, not nanoplastics. But the core finding is useful: plastic paint on a cap is still plastic touching your drink. The closure counts.

Where “90,000 more particles a year” comes from

You’ve probably seen the headline that bottled water drinkers swallow 90,000 more plastic particles a year. It comes from a 2025 review of more than 141 studies in the Journal of Hazardous Materials, which reports “bottled water consumers ingesting up to 90,000 more particles than tap water consumers” (Sajedi et al.).

The number is older than the review. It matches a 2019 estimate by Cox and colleagues, who pooled data from 26 studies. People who meet their water needs “through only bottled sources may be ingesting an additional 90000 microplastics annually, compared to 4000 microplastics for those who consume only tap water” (Cox et al.). The review’s other headline figure, 39,000 to 52,000 particles a year, is Cox’s estimate for food. It’s a model, not a measurement, built from studies published before the nanoplastic counts above. Cox’s team said the estimates carry “large amounts of variation” and “are likely underestimates.” Treat 90,000 as a rough, probably low ballpark.

Tap water isn’t clean either

A 2018 tap water survey, also done with Orb Media, analyzed 159 samples from 14 countries and found human-made particles in 81% of them (Kosuth et al.). Nearly all were fibers, averaging 5.45 particles per liter. The United States came out worst: its 33 samples averaged 9.24 per liter, the highest mean of any country (Kosuth et al., Table 4). Particles were picked out by eye, never chemically confirmed, so the authors call them “anthropogenic debris” rather than microplastics.

Newer research is bigger but messier. A 2025 meta-analysis of 43 tap water studies put the pooled average at 56.98 particles per liter, with polyethylene, PET and polypropylene most common. Its authors warn that without a standard method, results are “inconsistent and sometimes contradictory,” which makes comparisons unreliable (Malekzadeh et al.).

One of the cleaner recent datasets comes from Amsterdam. Researchers measured six common plastics by mass with a chemical method (pyrolysis GC-MS). River and polder source water averaged around 48 to 51 micrograms per liter. Treatment removed 97–98% of it, and household taps averaged 0.21 micrograms per liter (Sefiloglu et al.). That’s one well-run system, not a promise about yours.

So which is worse, bottled or tap?

Most studies can’t be compared directly: different particle sizes and methods can move a count by orders of magnitude. (Kosuth’s three bottled samples averaged 3.57 particles per liter, below the tap average, but three samples prove nothing.)

The best head-to-head so far, one lab and one method, found bottled water had significantly more micro- and nanoplastics than treated tap water (Jamison Hart and Lenhart). It’s small: four plants near Lake Erie and six brands (Ohio State). But it points the same way as the rest of the evidence: bottled water comes with a bottle and cap that shed, and tap water skips them. And by WHO’s account, “Conventional drinking-water treatment can remove particles smaller than a micrometre” (WHO).

Where your tap water is safe to drink, bottled water is the worse bet. You’re paying a company a markup to add packaging, and the packaging is part of the problem.

Heat makes plastic shed: the baby bottle study

Researchers found that polypropylene baby feeding bottles “release microplastics with values as high as 16,200,000 particles per litre.” Sterilizing them and exposing them to high-temperature water “significantly increase microplastic release.” They estimated infants could be exposed to between 14,600 and 4,550,000 particles per day, depending on region. The authors didn’t claim harm; they called for “an urgent need to assess” it (Li et al.). The lesson travels: hot water and plastic make a bad mix.

Boiling: weirdly, it helps

For once, an old habit holds up. Boiling makes nano- and microplastics stick to the calcium carbonate that forms as limescale. “Boiling hard water (>120 mg L–1 of CaCO3) can remove at least 80% of polystyrene, polyethylene, and polypropylene NMPs size between 0.1 and 150 μm” (Yu et al.). The catch is in that sentence: hard water, three plastics. If your water is soft, expect less. And the plastic only leaves your glass if the scale stays in the kettle.

Filters: what the evidence actually supports

Filter marketing is loud; the evidence is thinner.

• Membranes beat carbon alone. In one study, two pitchers with microfiltration membranes removed 78–86% of spiked PVC fragments and 94–100% of PET fragments (30 to 1,000 micrometers). A carbon-and-ion-exchange pitcher put out more particles than went in (Cherian et al.). Those particles were large; nanoplastics weren’t tested.
• Home treatment units can make things worse. In Dogonbadan, Iran, household units put out more microplastics (20 per liter on average) than went in (11.66). (Cheriki et al.).
• “Certified for microplastics” means something narrower than it sounds. The NSF/ANSI 401 claim requires a filter to cut particles of 0.5 to under 1 micrometer by 85% (NSF’s Rick Andrew). The test particles “are derived from ISO fine test dust,” not plastic (Andrew).
• Reverse osmosis: we found no peer-reviewed study of household reverse osmosis against microplastics in real-world use, so we won’t repeat the “99.9%” claims online. One irony: the Columbia team suggested the polyamide in their bottled water could come from polyamide-based reverse osmosis membranes. That is their hypothesis, not a finding (Qian et al. reply).

What regulators are doing

The WHO reviewed the evidence in 2019 (WHO). Its conclusion: “Based on the limited information we have, microplastics in drinking water don’t appear to pose a health risk at current levels. But we need to find out more.” Absorption of the smallest particles “may, however, be higher, although the data is extremely limited” (WHO). The nanoplastics it flagged are exactly what the new bottled-water studies count.

California has gone further. A 2018 law, SB 1422, ordered the State Water Board to define microplastics in drinking water and require four years of testing and public reporting (State Water Board). In September 2021 the Board found “insufficient evidence at the time” to set a notification level, and a February 2025 timeline showed Phase I monitoring from fall 2023 to fall 2025 and Phase II from fall 2026 to fall 2028 (State Water Board, 2025). As of this writing, we couldn’t confirm that Phase II has begun or that Phase I results are out.

What we don’t know

• Whether drinking-water levels harm people. No study has shown that the plastic in drinking water, at the levels people drink, causes disease. The human evidence so far comes from plastic found in blood, arteries and organs, and it shows associations, not cause. We cover it in Microplastics are inside you right now.
• How much is really there. Counts depend on the smallest particle a method can see, and plastic is in lab air, lab water and lab gear. That’s what the fight over the Columbia count is about (Materić).
• What’s in the particles. Plastic carries additives. Particle counts don’t tell you what chemicals come along.

What to actually do, ranked

1. Drink tap water where it’s safe. Check your utility’s water quality report. The best direct comparison found more plastic in the bottle (Jamison Hart and Lenhart).
2. If you buy bottled water anyway, keep it cool. Don’t leave it in a hot car or in the sun; heat plus jostling drove the biggest releases in the Nebraska tests. And don’t keep refilling and reopening single-use bottles: the bottleneck and cap wear with every twist (Winkler et al.).
3. Never heat water, formula or food in plastic. Heat and sterilization drove the biggest releases in the baby bottle study. Use a glass or steel kettle.
4. Carry glass or stainless steel, and mind the lid. A bottle can’t shed plastic it isn’t made of, but plastic lids and painted caps still touch your water (ANSES). Keep lids out of the heat. Switching from plastic-bottled to glass-bottled drinks isn’t automatically a fix.
5. If your water is hard, boiling is a free bonus. Let the scale settle and leave it behind (Yu et al.).
6. If you buy a filter, pay for a physical barrier. A membrane did far better than carbon alone in lab tests. Look for NSF/ANSI 401 microplastics certification, and know what it tests.
7. Make it a policy problem, not a shopping problem. You can’t filter your way out of a plastic economy. Ask your utility whether it tests for microplastics. Back rules that cut plastic production at the source, and that make PepsiCo, Coca-Cola, Danone and Nestlé answer for what their packaging puts in the product.