Oxybenzone Sunscreen: What the FDA's Own Trials Found

When the FDAFood and Drug Administration's own scientists ran a maximal-use clinical trial of an oxybenzone sunscreen in 2019, then a larger one in 2020, they found that within two hours of a single application, 100% of participants exceeded the FDA's own threshold for additional safety studies — Matta 2020, n=24/24 of participants had oxybenzone in their bloodstream above the agency's own threshold for triggering further safety studies. After three weeks, 96% still above the threshold three weeks after a single application phase — Matta 2020, n=22/23 of those participants were still above that threshold. No other sunscreen ingredient in the trial behaved this way.

This is not an article about whether to wear sunscreen. UVUltraviolet radiation is an IARCInternational Agency for Research on Cancer Group 1 carcinogen, it causes skin cancer, the data is settled, and sun protection works. This is an article about which type of UV filter you choose. Mineral filters — zinc oxide and titanium dioxide — sit on top of the skin and are barely absorbed. Chemical filters absorb UV inside the skin, and the most-studied of them, oxybenzone, also absorbs into your bloodstream within hours of application and stays there for weeks. It's part of the broader endocrine-disruptor story this site keeps coming back to, and the evidence-base sits at an unusual point: the absorption is documented by the FDA's own pharmacology, while the human clinical-outcome studies don't yet exist. The eso-friendly position is the same one the FDA proposed in 2019 — only zinc oxide and titanium dioxide have the safety data to clear the bar.

Oxybenzone (C₁₄H₁₂O₃, also called benzophenone-3 or BP-3, CASChemical Abstracts Service 131-57-7) is a synthetic organic compound that absorbs ultraviolet light across the UVB band and the shorter end of the UVA band. It has been used as a chemical UV filterA class of sunscreen ingredients that absorb ultraviolet radiation by undergoing molecular excitation, then re-emit the energy as small amounts of heat. Distinct from mineral filters, which scatter and reflect UV rather than absorbing it. in sunscreens since the 1970s and was approved by the FDA for over-the-counter sunscreen use in 1978.

It also has a quieter second life. Oxybenzone is photochemically stable — meaning it doesn't break down quickly in light — which makes it useful as a UV stabilizer in products that aren't sunscreen at all: lipsticks, foundations, nail polishes, hairsprays, fragrances. In those products it isn't there to protect your skin. It's there to keep the colour from fading and the fragrance from oxidising in the bottle. The consumer encounters it both ways.

96.8% of NHANES participants had detectable urinary oxybenzone — Calafat 2008, n=2,517 of urine samples in the 2003-2004 NHANESNational Health and Nutrition Examination Survey cycle contained detectable oxybenzone, geometric mean 22.9 µg/L, in a representative sample of 2,517 Americans aged 6 and older Calafat et al. 2008. The detection rate didn't fall over the next decade — Han and colleagues' 2016 follow-up across five NHANES cycles found no statistically significant change in BP-3 concentrations from 2003 to 2012, with an increasing trend since 2005-2006. Han et al. 2016.

Two things matter about that table. The first is that the sunscreen route is not the only one — even people who never use a chemical sunscreen accumulate oxybenzone through cosmetics that contain it for non-SPF reasons. The second is that the exposure isn't seasonal. The picture in your head when you think 'sunscreen' is probably a beach. The picture in your urine sample is a year-round daily-cosmetic exposure. That's the gap the biomonitoring keeps catching.

The agency that approved oxybenzone in 1978 ran the first proper pharmacokinetic trial of it forty-one years later. Matta and colleagues published a pilot RCTRandomized Controlled Trial in JAMA in 2019 (n=24, four sunscreen formulations) and a larger follow-up in 2020 (n=48, six chemical filters across four formulations). The protocols were the same in both: a 2 mg/cm² application — the dose the FDA's labelling assumes consumers actually use — to 75% of body surface area, four times a day for four days, with thirty-plus blood samples per participant over the following three weeks Matta et al. 2019 Matta et al. 2020.

Here's what happened in the 2020 study. Participants arrived at a phase-1 clinical unit, had baseline blood drawn, and were randomized to one of four sunscreen formulations. A trained applicator coated 75% of their body surface — face, arms, legs, torso — with the assigned product, four times across the first day. Blood was drawn at fixed intervals: 1 hour, 2 hours, 4 hours, 6 hours, every few hours for the rest of day 1, then on days 2, 3, 4, 5, 7, 10, 14 and 21. Two hours after the very first application, every single oxybenzone participant — 24× of 24× — had plasma oxybenzone above the FDA's 0.5 ng/mLthe threshold above which the FDA recommends additional non-clinical toxicology studies, per its 2016 Maximal Usage Trial guidance — not a 'safety limit' but a trigger for further data threshold. Peak concentrations reached a geometric mean of 258.1 ng/mL for the lotion formulation and 180.1 ng/mL for the aerosol spray. Geometric means — i.e., the typical participant — not outliers.

Then came the part nobody expected. After day 4, no more sunscreen was applied. Most of the chemical filters cleared from the bloodstream over the following week or two — avobenzone, octocrylene, octisalate, and octinoxate were all back below the threshold by day 7 or day 10. Oxybenzone wasn't. At day 21 — three weeks after the participants had stopped using it — 96% of them (22× of 23×) still had plasma oxybenzone above the threshold. Homosalate also persisted, more weakly, at 55%. None of the others did. Three weeks is long enough that the sunscreen you put on for a beach day in early August would still be circulating when school started in September.

The 0.5 ng/mL threshold deserves its own paragraph because it is widely misframed. It is not a 'safety limit.' It comes from the FDA's 2016 nonbinding guidance for industry on maximal-usage trials of topical drug products — the level above which the agency recommends additional non-clinical toxicology studies (genotoxicity, reproductive, carcinogenicity, developmental). It's the line at which the regulator says 'we now need more data before we can stand behind this.' Crossing it isn't the same as proven harm. It is, however, the agency's own line — and the typical Matta 2020 lotion participant crossed it by a factor of about 516× the FDA threshold 258 ng/mL ÷ 0.5 ng/mL.

The blood-level findings sit in front of a longer evidence trail of low-dose exposure routes nobody had been measuring. Oxybenzone has been detected in 85% of Swiss breast milk samples in a 54×-pair mother-child cohort Schlumpf et al. 2010. Krause and colleagues' 2018 study of paired maternal-fetal samples — 200 pregnant women paired with amniotic fluid plus a small cordocentesis subset — found BP-3 in amniotic fluid and cord blood at roughly an order of magnitude lower than in maternal serum Krause et al. 2018. The placental barrier reduces transfer; it does not prevent it.

There is no published human clinical-outcome study showing that oxybenzone exposure at consumer doses causes a specific disease. That is the honest framing. The case for treating oxybenzone as a chemical worth avoiding rests on four things instead: the bloodstream absorption documented above; the near-universal biomonitoring detection from Calafat onwards; in vitro and rodent evidence of endocrine activity; and the regulatory direction of travel in the EU, the SCCS, and the Pacific island jurisdictions that depend on the product class for tourism.

The endocrine signal in laboratory and animal work is not subtle. Oxybenzone binds estrogen receptors in cell-based assays, and Wnuk and colleagues' 2019 mouse study showed that prenatal exposure altered autophagy and disrupted RXR/PPAR-γ signalling in offspring brain neurons Wnuk et al. 2019. RXRs/PPARγRetinoid X receptors and peroxisome proliferator-activated receptor gamma — nuclear receptors that regulate gene expression for metabolism, neurodevelopment, and inflammation. Disruption is a recognized endocrine-disruption mechanism. are exactly the kind of molecular pathway you'd expect a hormone-mimicking chemical to perturb. Krause and colleagues' 2012 review across the broader UV-filter class flagged a consistent pattern of estrogenic, anti-androgenic, and anti-thyroid activity in animal models at doses overlapping with consumer biomonitoring ranges Krause et al. 2012.

The human epidemiology is thinner. Mustieles and colleagues' 2023 review of the toxicological and human evidence — the most recent comprehensive synthesis — pooled biomonitoring data and birth-cohort findings and concluded the human data on specific outcomes (gestational age, birth weight, child neurodevelopment) is suggestive but not consistent enough to establish causation Mustieles et al. 2023. The EUEuropean Union SCCSScientific Committee on Consumer Safety reached a similar verdict in its 2021 final opinion: the endocrine-disruption evidence on BP-3 was, in the committee's words, 'inconclusive and at best equivocal.' That is a regulator saying: the lab signal is real, the human studies don't yet pin down a clinical consequence, and we are reducing the maximum allowed concentration anyway. SCCS/1625/20 final opinion, adopted 30-31 March 2021.

It is also worth saying what the dermatology profession says, because there's a real argument from that side. The American Academy of Dermatology's position is that there is no scientific evidence sunscreens cause adverse health effects in humans, that the absorption findings do not establish harm, and that the most important thing is for people to keep using sunscreen — whichever kind they prefer — because skin cancer kills people the FDA's threshold doesn't. The Matta authors themselves wrote, in both 2019 and 2020, that 'these findings do not indicate that individuals should refrain from the use of sunscreen.' That sentence is in the discussion section of every paper this article has cited so far. It is not a throwaway.

The paper that pushed oxybenzone into the public conversation as a coral-reef threat came in 2016. A team led from Haereticus Environmental Laboratory collected reef seawater samples from snorkelling beaches in the U.S. Virgin Islands and Hawaii — popular tourist sites where dozens of swimmers were rinsing sunscreen off into the water every hour — and measured BP-3 concentrations of 75 µg/L to 1.4 mg/L at the highest USVI sites and 0.8-19.2 µg/L in Hawaii. They then tested oxybenzone toxicity on coral planulae and cultured coral cells from seven species under aquarium-grade UV light to mimic reef conditions. The dose that produced developmental abnormalities in coral planulae over 24 hours came in at 6.5 µg/L. The dose that killed 20% of cultured coral cells in four hours, in light, ran lower still — to 62 parts per trillionthe lower end of the range across seven coral species — Downs 2016, 4-hour LC20 in light, range 0.062-8 µg/L at the bottom of the seven-species range. Sixty-two parts per trillion is roughly three drops of oxybenzone in an Olympic-sized swimming pool Downs et al. 2016.

The Pacific island and Caribbean jurisdictions read the paper, looked at the snorkelling-beach concentrations, and acted. Hawaii passed Act 104 in 2018 banning the in-state sale of sunscreens containing oxybenzone or octinoxate from 2021. Palau's Responsible Tourism Education Act took effect 2020 with a list of ten banned ingredients. The U.S. Virgin Islands followed in March 2020, becoming the first jurisdiction globally to also ban octocrylene. Bonaire, Aruba, and Maui County have similar restrictions. The precautionary principleA regulatory approach treating plausible serious harm with available alternatives as grounds for action, even where the human-clinical evidence base is incomplete. is doing real work here — none of these regulators are claiming the lab study has been replicated at whole-reef scale.

A 2021 critical review pointed at exactly that gap. Industry-adjacent and methodologically sceptical, the analysis noted that the USVI seawater limit-of-quantitation was fifty times the limit-of-detection, that only 25% of the BP-3 coral-toxicity studies up to that point reported method blanks, and that field-blanks were reported in only one study. None of which proves the lab work is wrong; it tells you the field-evidence quality control could be better than it has been Mitchelmore et al. 2021. The honest summary: lab and ex vivo coral toxicity at low parts-per-trillion is real, the regulatory response is precautionary, and the whole-reef field evidence is messier than the headlines suggested. The case for switching does not rest only on the corals — they're a smaller part of the story than the human bloodstream data.

The regulatory picture splits along a familiar line. The EU has moved on the cosmetics side. The US has moved on the local cosmetics side and stalled federally. The Pacific has moved on the environmental side. Nobody has banned oxybenzone outright in a major Western market — every restriction is concentration-capped or product-category-specific.

On the federal US side: the FDA's February 2019 Proposed Rule (84 FR 6204) classified twelve sunscreen actives — oxybenzone, octinoxate, homosalate, avobenzone, octocrylene, octisalate, cinoxate, dioxybenzone, ensulizole, meradimate, padimate O, and sulisobenzone — as Category III, meaning 'insufficient data to determine generally recognized as safe and effective.' Only zinc oxide and titanium dioxide were classified as Category I — GRASEGenerally Recognized as Safe and Effective. PABAPara-aminobenzoic acid — an early UV-B filter widely used through the 1980s and now mostly removed from sunscreens because of high allergic-contact dermatitis rates. and trolamine salicylate were classified Category II (non-GRASE). Under the CARES Act of 2020, the proposed rule was rolled into a September 2021 proposed administrative order, which has not been finalized as of April 2026. Existing oxybenzone-containing products remain legally marketable in the United States under a deemed-final-order interim status.

On the EU side: Commission Regulation (EU) 2022/1176, adopted 7 July 2022, amended Annex VI of the Cosmetics Regulation 1223/2009 and cut the maximum permitted oxybenzone concentration to 6% in face creams, hand creams, and lipsticks; 2.2% in body creams and pump or aerosol sprays; and 0.5% as a non-UV-filter formulation stabilizer. The EU Cap 2.2% body / 6% face 2023 limits applied universally from 28 July 2023. The reduction was driven by the SCCS/1625/20 opinion; the SCCS itself characterized the endocrine-disruption evidence as inconclusive but recommended the cut anyway, on the basis of the MoSMargin of Safety calculation at the prior 6% body-product limit being too low under realistic-use exposure modelling.

Hawaii's situation is more layered than 'Hawaii banned sunscreen' suggests. Act 104 of 2018 banned the sale of sunscreens containing oxybenzone or octinoxate from 1 January 2021, with a prescription exception. A subsequent attempt to add avobenzone and octocrylene to the state ban (SB 132) passed the Senate but died in House committee — the state-level restriction remains the original two filters. Maui County Ordinance 5306 went further at the county level, banning all non-mineral sunscreens (everything except zinc oxide and titanium dioxide) effective 1 October 2022. Key West, Florida tried a similar municipal ban in 2019, but Florida SB 172 (2020) preempted local cosmetics regulation before the Key West ordinance took effect.

If you have decided the precautionary case is strong enough — which is a personal call, not a regulatory one — the practical move is two-step. Read the active-ingredients box on the back of any sunscreen, lip balm with SPF, BB cream, or tinted moisturizer with SPF. If 'oxybenzone' or 'benzophenone-3' is listed, the product contains it. Then check the front label for 'mineral' or 'physical' sunscreen — those products use zinc oxide or titanium dioxide as the active. The Switch is straightforward; mineral formulations now exist with cosmetic elegance comparable to chemical ones, often with tinted versions that solve the white-cast problem.

What this article does not say is 'stop wearing sunscreen.' UV protection prevents skin cancer; that part of the evidence has been settled for a long time, and the dermatology profession has been right to keep saying it. The argument is narrower: if a chemical filter and a mineral filter give you the same UV protection, but the chemical filter is in your bloodstream above the FDA's own guidance threshold for three weeks after a single application — and the mineral filter sits inert on top of your skin — the precautionary calculus favours the mineral one, and the cost of the switch is essentially nothing.

Mineral filters are the eso-friendly answer here, and there is unusually clean regulatory consensus to support that: zinc oxide and titanium dioxide are the only two sunscreen actives the FDA has classified as Category I, generally recognized as safe and effective, in its 2019 proposed rule. Both have minimal systemic absorption through intact skin. Both protect across the full UVA-UVB range when properly formulated. Look for products where one or both is the only listed active ingredient — and be wary of 'mineral-based' marketing that hides chemical filters in the ingredients list. The back-label active-ingredients box is the only thing that matters.

Forty-one years after oxybenzone was approved for over-the-counter sunscreen use, the agency that approved it ran the first proper pharmacokinetic trial of how much of it ends up in the bloodstream. The answer was: a lot, fast, and for longer than anyone expected. The FDA's own 0.5 ng/mLthe threshold above which the FDA recommends additional non-clinical safety studies-threshold-for-further-study line was crossed by every participant within two hours of a single application; for the lotion arm the typical participant crossed it by a factor of about 500; and three weeks after the dosing stopped, almost everyone was still over the line. In a regulatory system designed to surface signals like that, the rest happens slowly. The proposed rule that would settle the question has been pending since 2019. The EU has moved its limits twice since then and the SCCS has hedged its endocrine verdict while recommending lower caps — the same hedged-but-restricting pattern the EU took on parabens. The mineral alternatives were already on the shelf, made by the same companies, in the same aisles, at the same price.