Evidence by System

Homes and car cabins, Toulouse (Yakovenko et al., 2025). The team sampled indoor air in three apartments and two cars from January to May 2023 (12 samples) and identified particles by Raman spectroscopy down to about 1 µm. Median concentrations were 528 particles/m³ in homes and 2,238 particles/m³ in cars, and 94% of particles were 1–10 µm, small enough to reach the deep lung. With published data, the authors estimate that an adult inhales about 68,000 such particles a day, an estimate, not a measurement. Limitation: only 0.3% of each filter was analyzed and scaled up, and the home–car difference was not statistically significant.

Size-fractionated roadside air, Leipzig (Kaushik et al., 2025). Particulate matter below 10 µm was sampled at one busy roadside site for two weeks in September 2022 and analyzed by pyrolysis gas chromatography–mass spectrometry (py-GC/MS), which reports polymer mass rather than particle counts. Plastic averaged 0.6 ± 0.2 µg/m³, split evenly between fine and coarse fractions. Tire wear particles made up about 65% of it. The polyvinyl chloride (PVC) signal cannot be told apart from soot. Limitation: one site, which the authors treat as an upper-limit exposure; their mortality risk figures apply general fine-particulate coefficients to plastic and are model estimates, not observed outcomes.

Oslo and remote Norway (Herzke et al., 2026). Suspended and deposited airborne microplastics were sampled in parallel at urban and remote Norwegian sites and analyzed polymer by polymer with py-GC/MS. Microplastics appeared in nearly all samples. In Oslo, tire wear particles made up more than 90% of airborne microplastic mass and peaked when drivers switched to winter tires. In short episodes they reached up to 30% of fine particulate matter (PM2.5). Remote sites showed lower but steady levels, which the authors attribute to long-range transport from European source regions. Limitation: one country, and attribution at remote sites depends on transport modeling.

Cloud water above Japan (Wang et al., 2023). Researchers collected cloud water at three mountain sites between about 1,300 and 3,776 m altitude, including the summit of Mount Fuji, and identified particles by infrared imaging and micro-FTIR spectroscopy. They found nine polymer types at mean concentrations of 6.7 to 13.9 pieces per liter, with sizes from 7.1 to 94.6 µm. Many particles carried water-attracting groups, which the authors propose could help seed droplets or ice. It shows plastic traveling inside clouds. Limitation: the size floor near 7 µm misses nanoplastics, and the cloud-seeding role is a hypothesis, not a measurement.

Gloves as a source of false positives (Clough et al., 2026). While measuring airborne microplastics in Michigan, the team traced spurious counts to stearate salts on disposable gloves. Their infrared and Raman spectra resemble those of microplastics, so library matching misidentifies them. Dry contact with ordinary nitrile and latex gloves left a mean of about 2,000 false positives per square millimeter; a nitrile cleanroom glove left about 100. The errors concentrate below 10 µm. Limitation: how many earlier datasets are affected is not established; the authors give workflows and stearate spectra to separate the two.

Floating plastic in the world's oceans, 1979–2019 (Eriksen et al., 2023). The team combined published and new surface-net data from 11,777 stations into a global time series. For 2019 they estimate 82 to 358 trillion floating plastic particles weighing 1.1 to 4.9 million metric tons. They find no clear trend before 1990, a fluctuating plateau until 2005 and a rapid rise since. Limitation: one modeled analysis of pooled data, not yet independently replicated; stations are unevenly spread, and surface nets miss the smallest particles.

Where buoyant ocean plastic sits (Kaandorp et al., 2023). A three-dimensional mass-budget model assimilated observations from coastlines, the ocean surface and the deep ocean for pieces from 0.1 to 1,600 mm. It estimates 3,200 thousand metric tons (kt) of initially buoyant plastic in the ocean in 2020, more than 95% of it (3,100 kt) in pieces larger than 25 mm. Input is about 500 kt a year. On these numbers there is no "missing" ocean plastic. Limitation: a model estimate covering only initially buoyant plastic; nanoplastics are not resolved.

Lakes and reservoirs in a standardized survey (Nava et al., 2023). A cross-national team sampled surface water in 38 lakes and reservoirs for plastic debris larger than 250 µm, using one protocol everywhere. Every water body contained plastic. The most affected were lakes in densely populated, urbanized areas and large lakes with long water-retention times; in the most polluted, concentrations reached or exceeded those reported in subtropical ocean gyres. Limitation: one consortium study; surface water only, and the 250 µm floor excludes small microplastics and all nanoplastics.

Nanoplastic mass in Australian drinking water (Shi et al., 2026). The team screened tap and bottled water by py-GC/MS, separated particles by size with field-flow fractionation, and checked results with particle tracking and electron microscopy. Average concentrations were 0.3 to 0.7 µg/L for nanoplastics (10–1,000 nm) and 0.5 to 1.5 µg/L for selected microplastics, higher in tap than in bottled water. It combines several independent methods; two of its authors also wrote the blood method study below. Limitation: the polypropylene identification in the nano fraction is called indicative, and tap-water particle counts (one to ten million per milliliter) are not polymer-specific.

Thirty years of plastic mulch in Xinjiang cotton fields (Zhang et al., 2025). The team sampled 60 field points with 5 to 30 years of mulching history at three depths down to 30 cm. Microplastics of 20–500 µm were extracted by density separation and confirmed by infrared spectroscopy. After 30 years, residual film totaled about 281 kg/ha in the top 30 cm, and the top 10 cm held about 10,000 to 10,800 microplastic fragments per kilogram (the paper gives both values). Accumulation sped up once residual film passed about 160–200 kg/ha. Limitation: one arid region, different fields standing in for different ages, and particles below 20 µm excluded.

Nanoplastics in a compost-amended field, Loire valley, France (Ryzhenko et al., 2026). The field received large amounts of household-waste compost in the 1990s. Soil cores to 90 cm from an amended and a reference plot were composited per plot; the 200–1,000 nm fraction was isolated by filtration and polystyrene (PS) measured by matrix-matched py-GC/MS. Topsoil (0–5 cm) held 2.57 µg/g dry weight, about 28 times the reference plot (0.09 µg/g). Grass roots held 123 µg/g, not size-selected; shoot values were below the quantification limit. Limitation: one site, one polymer, one composite sample per plot; recovery and blank data sit in a supplement we did not read.

Nanoplastics from roots to pollen (Piovesan et al., 2026). Pumpkin plants were exposed to radiolabeled polystyrene (PS) nanoplastics of about 20 nm at 1 µg/L, near expected environmental levels, and at 1 mg/L, a worst case. The label moved from roots to flowers and concentrated in the anthers; a radiolabel tracks the carbon, not necessarily intact particles. The authors report lower pollen viability, abnormal pollen shape and hydration, and reduced photosystem efficiency in leaves. Limitation: one crop, laboratory conditions, uniform PS spheres; we read the abstract only, which does not say which effects occurred at the lower dose.

A global estimate of photosynthesis and crop losses (Zhu et al., 2025). A meta-analysis of 3,286 data points reports that microplastic exposure lowers photosynthesis by 7.05 to 12.12% across land plants and marine and freshwater algae. A model then estimates annual losses of 4.11 to 13.52% of main crop production, or about 110 to 361 Mt. The crop figure is disputed. Two experts quoted by the UK Science Media Centre call it speculative, since slower photosynthesis may delay growth without lowering final yield; a third sees the study as added evidence for action²⁸. Limitation: an extrapolation from exposure experiments to field conditions worldwide.

Placenta mass by py-GC/MS (Garcia et al., 2024). The team digested 62 placentas, isolated solids by ultracentrifugation and quantified 12 polymers by py-GC/MS. Every sample contained plastic, from 6.5 to 685 µg per gram of tissue (mean 126.8 µg/g), and polyethylene was 54% of the total. Fluorescence microscopy showed particles and fibers above 1 µm but no nano-sized particles. The amount is disputed: critics of the same laboratory's brain workflow cite fat interference and missing validation²⁹, and in blood py-GC/MS gave false polyethylene signals¹⁹; the placenta itself was not retested. Limitation: no independent confirmation of the amounts.

Olfactory bulb, São Paulo (Amato-Lourenço et al., 2024). Olfactory bulbs from 15 deceased adults who had lived in São Paulo for more than five years were examined by micro-FTIR spectroscopy. Microplastics were found in 8 of the 15: 16 particles and fibers in total, 44% of them polypropylene (PP), with particles 5.5 to 26.4 µm across. Procedural blanks contained no polymers. The olfactory bulb is a possible route into the brain that bypasses the blood–brain barrier. Limitation: 16 particles from 15 people, none reported below 5 µm, and the route itself was not shown.

Urine of 10-year-olds, Ma'anshan Birth Cohort, China (Liu et al., 2026). Urine from 1,308 children was screened by laser direct infrared imaging, whose detection floor is 20 µm. Microplastics were detected in 91% of samples at a median of 250 particles/mL; 98% of detected particles were 20–100 µm, and polytetrafluoroethylene (PTFE) appeared in half of the samples. Diet quality was associated with the abundance of some polymer types. Limitation: we read the abstract only, so collection and blank controls are unchecked. Particles this large exceed what healthy kidneys filter, and so high a count makes contamination a question to settle first (our note).

Brain tumors and healthy brains, China (Li et al., 2026). The team analyzed 156 samples from 113 patients having brain-tumor surgery and 35 samples from eight regions of five post-mortem donors. Particles were counted by laser direct infrared imaging, a subset of samples went to py-GC/MS, and single particles were confirmed by optical photothermal infrared spectroscopy. Surgical air, syringes, infusion sets and other items were tested as contamination sources. Plastic was detected in 99.4% of patient samples and all donor samples. In 23 samples from the five donors, the py-GC/MS median was 50.3 µg/g, a value we take from the publisher's source-data file, about one hundredth of the Nihart 2024 median (our comparison)³⁰. Limitation: samples are not independent people, and the design cannot show cause; a correction fixed figure rounding only³¹.

What the health studies add up to (Chartres et al., 2024). A rapid systematic review using the Navigation Guide method searched four databases in July 2022 and April 2024. It included 3 human observational studies and 28 animal studies on digestive, reproductive and respiratory outcomes. The authors conclude that microplastics are "suspected" to harm these three systems, rating some bodies of evidence high quality, and call birth outcomes "unclassifiable". Limitation: the human evidence rests on three observational studies; most findings come from animals at the tested doses.

Testing the main mass method in blood (Rauert et al., 2025). The team developed and validated a py-GC/MS extraction for human blood, then ran a pilot with eight Australian adults sampled fasting and after a meal. Recoveries ranged from 7 to 109%, and realistic detection limits were up to 20 times higher than limits measured in clean water. Polyethylene signals from the blood itself were reduced but not removed, and no other polymer exceeded detection limits. The authors judge py-GC/MS currently unsuitable for polyethylene and PVC in biological matrices, and for PET in blood because of matrix suppression. Limitation: blood only, one laboratory.

Confidence levels for plastics in human samples (Thomas et al., 2026). This framework paper ranks analytical techniques by how specifically they identify plastic and sets minimum data from independent, orthogonal techniques for each confidence level. It asks for transparent reporting of contamination controls and uncertainty before amounts in human tissue inform health decisions. An independent critique of the brain measurements raises the same concern²⁹; the blood method study shares two authors with this paper¹⁹. Limitation: a proposal, not a validated or adopted standard.