Don’t take our word for it. Here’s every fact on this site, with its evidence grade, its caveats, and a link to the source, so you can check us.
We start you on the strong ones: findings that have been replicated or come from one large, well-controlled study. Moderate and weak facts are here too, labeled for what they are. Use “Link to this fact” to share a single card.
Indoor air / Other evidence
The white musk pilot used only 10 microliters of undiluted oil on a swab in a 4.2 cubic meter booth, chosen because larger amounts (up to 100 microliters) smelled too strong, and never measured air concentrations of any compound.
Moderate evidence: Describes the design only; if all 10 uL evaporated, galaxolide would reach about 83 ug/m3 (my upper-bound arithmetic from the paper's numbers); actual exposure unknown.
Source: Zennifa 2025, Scientific Reports Industry-funded Link to this fact
Indoor air / Other evidence
In a mouse test, mixing the pine and citrus scent chemicals alpha-pinene or limonene with ozone made a gas that cut the animals' breathing rate by about 30% and 33%; each chemical alone, at similar levels, and the ozone alone did not.
Moderate evidence.
Caveats
Mouse study (ASTM bioassay, BALB/c, n=12 per condition) at about 50-60 ppm terpene and 4-6 ppm starting ozone, far above indoor levels; part of one lab group's work; the same group later attributed about 75% of the limonene effect to formaldehyde plus leftover limonene.
Source: Wolkoff 2000, Indoor Air Industry-funded Link to this fact
Indoor air / Other evidence
The concentrations in this mouse test were much higher than found indoors: about 48-56 ppm of leftover terpene and 4-6 ppm of starting ozone, versus indoor limonene or pinene maxima of roughly 45-70 ppb and indoor ozone up to roughly 150 ppb (the wiki's conversions of the paper's tables).
Moderate evidence.
Caveats
Our arithmetic from the paper's printed tables; the authors justify the doses by mouse insensitivity (about 10x less sensitive than humans) and the need for over 90% ozone consumption in 30 s.
Source: Wolkoff 2000, Indoor Air Industry-funded Link to this fact
Indoor air / Other evidence
In a controlled office experiment with 30 young women, more outdoor air (3, 10, then 30 liters per second per person) made the air seem fresher on entering (about 38% then 31% then 15% dissatisfied) and reduced dry mouth and throat, difficulty thinking clearly and feeling bad, while a carpet that gave off chemicals stayed in the room.
Moderate evidence.
Caveats
One 4.6-hour exposure per condition, 30 young female students, self-reported ratings, subjects could smell the difference on entering; percentages approximate (read from Fig. 4); the improvement in air quality during the sit was not significant.
Indoor air / Other evidence
Raising office ventilation did not only dilute the chemicals: higher outdoor air also brought ozone indoors (0, 4 and 16 ppb at 3, 10 and 30 L/s per person) and the room's measured sensory pollution load rose at the highest rate, so the authors favored removing sources plus ventilating rather than ventilation alone.
Moderate evidence.
Caveats
Single chamber-like office with a constant carpet source; no ozone-VOC secondary products were detected (aldehydes below detection limits), so the trade-off is a caution, not a measured harm.
Indoor air / Other evidence
A carpet used as the constant pollution source released about 127 micrograms of volatile chemicals per square meter per hour, including toluene, acetic acid, hexanal and small amounts of the scent terpenes limonene and alpha-pinene.
Moderate evidence: One used carpet, lab emission test at 23 C; terpenes were a small part of the mixture; not a fragrance product.
Indoor air / Other evidence
A 2005 critical review of 30 primary studies (from over 500 screened) found little strong research on whether classroom air quality changes student performance, and no pollutant-performance link it could call persuasive.
Moderate evidence: Narrative review of literature through July 2003, authors' own grading; later studies exist.
Indoor air / Other evidence
Lower ventilation rates were suggestively but not consistently linked to worse performance: in one office laboratory study each doubling of ventilation between 3 and 30 L/s per person raised task output by 1.1-2.1%, while another found no difference.
Moderate evidence: Adult office-laboratory data; Myhrvold 1996 in schools was a brief, non-peer-reviewed report; CO2 is only a ventilation marker here, so this is not a direct CO2 effect.
Indoor air / Other evidence
Office workers with lower ventilation (25 vs 50 cfm per person) had 1.5 times (1.2-1.9) the short-term sick leave in one strong study, though a second study at high ventilation levels found no association.
Moderate evidence: Adult offices, rough ventilation estimates; may not transfer to classrooms.
Indoor air / Other evidence
The review found no included study linking indoor particles or biological pollutants directly to performance or attendance; the evidence for dampness and mold runs through asthma, respiratory infection and absence instead.
Moderate evidence: Reflects the literature in 2003; indirect links were summarized but not critiqued.
Indoor air / Other evidence
In 100 southwestern US elementary schools, the fifth-grade classroom's estimated outdoor-air ventilation had a median of 3.55 liters per second per person, about half the 7.1 L/s minimum in the ASHRAE 2004 standard, and 87 of 100 classrooms were at or below that minimum.
Moderate evidence.
Caveats
Ventilation was inferred from one or a few days of peak CO2 with windows closed, uncertainty about 1 L/s per person; two districts in one state, so not nationally representative; part of the funding came from HVAC industry groups.
Source: Haverinen-Shaughnessy 2011, Indoor Air Link to this fact
Indoor air / Other evidence
In all 100 schools together, including those above the recommended ventilation rate, classroom ventilation was not significantly related to test scores (math slope 1.02, P=0.18; reading 1.31, P=0.16), and the fitted curve turned down above roughly 6 L/s (approx., read from Fig. 3), so the benefit of ventilation beyond the standard is unknown.
Moderate evidence: Only 13 schools were above 7.1 L/s so power is low; the authors say a larger sample is needed.
Source: Haverinen-Shaughnessy 2011, Indoor Air Link to this fact
Indoor air / Other evidence
Because the CO2 was injected as ultrapure gas with ventilation, temperature and VOCs unchanged, the study tests CO2 itself rather than the other pollutants that rise when ventilation is low.
Moderate evidence: This isolates CO2 only for this test and sample; it says nothing about VOC, fragrance or bioeffluent effects.
Source: Satish 2012, Environmental Health Perspectives Link to this fact
Indoor air / Other evidence
In a blinded 24-person office-lab study, scores on a complex decision-making simulation were about 61% higher on a low-VOC day (TVOC about 45 µg/m³) than on a day with about 590 µg/m³ TVOC from common office products (tape, markers, correction fluid, cleaner, wipes), and about twice as high on high-ventilation low-VOC days.
Moderate evidence.
Caveats
One test day per condition, fixed condition order, n = 24, one industry-gift funder (United Technologies), the VOC contrast rests on a single day with one set of added products; no fragrance or terpene measured.
Source: Allen 2016, Environmental Health Perspectives Industry-funded Link to this fact
Indoor air / Other evidence
When limonene reacted with ozone in the test chamber, formaldehyde and two other irritant-type chemicals (4-oxopentanal and 3-isopropenyl-6-oxoheptanal) formed within the first hour, at roughly 9-17 micrograms per cubic meter each, along with a burst of ultrafine particles (average about 830 per cm3, maximum about 2,200).
Moderate evidence: Chamber chemistry at 0.7 air changes/h and 22 C with dry air; one run per condition for the chemistry; levels the authors call the upper end of indoor air.
Source: Verstraelen 2024 (J Environ Sci Health A), Journal of Environmental Science and Health, Part A Link to this fact
Indoor air / Other evidence
In one public restroom, three 222 nm germicidal UV lamps added only about 4 to 6 ppb of ozone, but that ozone reacted with fragrance terpenes from urinal deodorizer screens and with surfaces, roughly doubling to tripling the ozone chemistry in the room.
Moderate evidence: One unoccupied 59 m³ restroom, two weekends, NIST authors, no health endpoint; not generalizable to other rooms.
Source: Link 2024, Environmental Science: Processes & Impacts Link to this fact
Indoor air / Other evidence
The lamp-driven chemistry made formaldehyde, acetaldehyde, hexanal, acetic acid, acetone and other oxidation products (34 in all), at a lamp-attributable rate of about 9 ppb per hour at 1 air change per hour and about 6 ppb per hour at 2.
Moderate evidence: Individual formaldehyde concentration not reported separately; variability 30 to over 50%; one room.
Source: Link 2024, Environmental Science: Processes & Impacts Link to this fact
Indoor air / Other evidence
The lamps also created new ultrafine particles, with average aerosol mass 1.8 micrograms per cubic meter (maximum 2.6) at 1 air change per hour; at 2 air changes per hour aerosol mass was undetectable but ultrafine particles still formed.
Moderate evidence: Particle size range 10-420 nm only; composition not measured; ventilation confounded with older screens.
Source: Link 2024, Environmental Science: Processes & Impacts Link to this fact
Indoor air / Other evidence
In a Boston-area study of 54 graduate students in classrooms with increased ventilation, higher peak classroom CO2 (95th percentile) was associated with slower Stroop and arithmetic responses, about 5-6% slower per 100 ppm, across a low range of roughly 440-1,630 ppm.
Moderate evidence.
Caveats
Association only; small, mostly young female sample at one well-funded university; CO2 reflects ventilation and occupancy, not a proven direct effect; no multiplicity correction; adjusted R2 <=0.25.
Source: Dedesko 2025, Journal of Exposure Science & Environmental Epidemiology Link to this fact
Indoor air / Other evidence
This classroom study cannot say whether VOCs affect thinking: it measured TVOC (mean 340 ppb) and PM2.5 but dropped them from the analysis because low-cost sensors were unreliable.
Moderate evidence: The authors only hypothesize that human-emitted VOCs vary with CO2; no VOC effect estimate, positive or null, is reported.
Source: Dedesko 2025, Journal of Exposure Science & Environmental Epidemiology Link to this fact
Indoor air / Other evidence
The study treats CO2 as a marker of how much fresh air a classroom gets and how many people are in it, not as proof that CO2 itself impairs thinking at these levels.
Moderate evidence: Ventilation rate and occupancy were not measured; demand-controlled ventilation makes rates dynamic; effects of pure CO2 injection are mixed in the wider literature.
Source: Dedesko 2025, Journal of Exposure Science & Environmental Epidemiology Link to this fact
Indoor air / Other evidence
In a single-blind chamber study of 71 university students, a room at 28 C produced more errors than one at 23 C on several memory and attention tests (for example about 7.6 percentage points more on a visual-memory task), while the room's air stayed at CO2 under 700 ppm.
Moderate evidence.
Caveats
One lab, 60 min exposure, university students, 23 vs 28 C only, BARS battery on laptop keyboard, no multiplicity correction, between-subject VOC groups; heat finding, not a chemical finding.
Source: Zhao 2025, Science of The Total Environment Link to this fact
Indoor air / Other evidence
In a randomized experiment with 398 healthy adults in a simulated aircraft cabin, about 2.75 hours of extra pure CO2 (measured up to about 4,100 ppm) did not produce a consistent drop in eight kinds of thinking tests.
Moderate evidence.
Caveats
Single study, short exposure, healthy adults, industry (aerospace + ASHRAE) funded; the factorial design collapsed so CO2 was never isolated from VOC and pressure; no effect sizes reported.
Source: Herbig 2026, International Journal of Hygiene and Environmental Health Industry-funded Link to this fact
Indoor air / Other evidence
'Formaldehyde-free' paint still released dozens of other volatile compounds: in one chamber test, trimethylbenzenes and xylenes ran 15 to 35 times above background, and no formaldehyde was detected.
Moderate evidence: One speciated sample of one paint brand in a chamber; not a measurement of homes.
Source: Sun 2026, Journal of Hazardous Materials Link to this fact
Indoor air / Other evidence
In a mouse pregnancy study, the VOC exposure (0.65 ppm over 3 hours, about 2.5 times China's 0.6 mg/m3 TVOC limit by the authors' conversion) was set by the authors above typical indoor levels, so the results show hazard at a high-end renovation-type dose, not risk at ordinary home levels.
Moderate evidence: Authors say it is below maximum post-renovation levels they cite; speciated table sums to about 0.9 mg/m3 rather than 1.5 mg/m3.
Source: Sun 2026, Journal of Hazardous Materials Link to this fact
Indoor air / Test-chamber measurement
In a 2.5 m3 test room ventilated at about 0.5-0.8 air changes per hour, pulses of 60-100 ppb ozone added to air containing limonene (the lemon scent in many air fresheners and polishes) produced new submicron particles within minutes; later pulses grew them to larger submicron sizes.
Moderate evidence: Single runs per condition, one lab, injected limonene, ozone set by a generator; particles below 0.1 um and particle chemistry were not measured; no health outcome.
Source: Wainman 2000, Environmental Health Perspectives Link to this fact
Indoor air / Other evidence
Indoor air is not automatically a refuge from ozone: calculations from 12 sets of published personal-monitor data put about 27-60% of people's daily inhaled outdoor-origin ozone indoors (average just over 40%), mostly because of the time spent there.
Moderate evidence.
Caveats
A single author's calculation from other groups' monitoring data, using total breathing rates and assumed activity levels; it is an intake estimate, not a measure of harm. Abstract (25%) and text (27%) differ slightly. No fragrance data.
Source: Weschler 2006, Environmental Health Perspectives Link to this fact
Indoor air / Other evidence
Soot (elemental carbon) measured inside classrooms was about 94% of the level outdoors, and higher indoor soot went with a 13% smaller one-year gain in working memory (4.2% to 23.1%) between the lowest and highest quarters of schools.
Moderate evidence: Single classroom per school, two 1-week measurements; observational association, not a test of filtration or ventilation.
Source: Sunyer 2015 (BREATHE, Barcelona), PLOS Medicine Link to this fact
Indoor air / Other evidence
A computer model of typical US homes finds indoor ozone reacts almost only with terpene scent chemicals: about 99% of its reactions with indoor VOCs, with limonene (citrus) about two-thirds and alpha-pinene (pine) about a quarter.
Moderate evidence: Model, not a measurement; VOC profile is a literature median (Logue 2011), not a home with heavy scented-product use; no validation against field data.
Source: Waring & Wells 2015, Atmospheric Environment Link to this fact
Indoor air / Other evidence
In 30 closed Korean homes with no other scented products allowed, putting three Vicks plants in the living room raised room limonene about 1.5 times for plants in soil and about 2.3 times for plants in water over 3 hours; alpha-pinene rose about 1.5 and 2.5 times.
Moderate evidence.
Caveats
One plant species, 3-hour windows, 30 homes, conditions not randomized in order (soil day then water day), small closed rooms; the alpha-pinene rise for soil plants was not significant (p = 0.057).
Indoor air / Other evidence
In a lab chamber with a very high ozone level (about 6.5 ppm), all six ordinary perfumes, deodorants and cleaning liquids tested formed new particles; about 0.2% to 17% of the product's evaporated mass became aerosol, with an average near 4%.
Moderate evidence.
Caveats
One lab, six products, a closed 0.68 m3 chamber at roughly 100 times typical indoor ozone, very dry air, no vapor wall-loss correction; a yield, not an indoor concentration. NSF and Sloan funded; no industry funding.
The industry / Self-reported survey
Even the companies selling products often don't know everything in them: all 20 'leader' consumer-product companies interviewed said ingredient or hazard information could be hard to get from suppliers, and 7 named suppliers' trade-secret claims as a major obstacle.
Moderate evidence: Interview study, 20 self-selected companies, 2009; no fragrance firms identified.
Source: Scruggs 2011, Environmental Science & Policy Link to this fact
The industry / Self-reported survey
Only 1 of 20 leading consumer-product companies could test its own products' chemistry in-house; most relied on suppliers' signed promises (18 of 20).
Moderate evidence: Counts are interview mentions.
Source: Scruggs 2011, Environmental Science & Policy Link to this fact
The industry / Lab study (cells or chemistry)
The fragrance industry's own data show its main explanation for how fragrance chemicals cause allergy doesn't cover most of them: 34 of 62 ingredients (55%) sensitize despite having no predicted protein-binding alert.
Moderate evidence.
Caveats
Structural-alert software limitation; metabolic activation explains some. Roberts & Api 2018 (different 57-chemical fragrance dataset, no direct overlap confirmed with this claim's 62-chemical set) adds two further indirect-activation routes the alert software here evidently does not capture — slow autoxidation of tertiary allylic/benzylic C-H compounds to allergenic hydroperoxides/epoxides, and a proposed bio-sulfation route for benzylic alcohols — which could explain part of the unexplained 55% for chemicals with those structural motifs. The same paper separately proposes that some LLNA-positive, alert-free fragrance chemicals (hexyl/benzyl salicylate) may be LLNA false positives rather than true sensitizers at all, which would pull in the opposite direction (shrinking, not explaining, the true no-mechanism fraction) if any such chemicals are present in this claim's 62-ingredient set — not checked here, so the net effect on the 55% figure is unresolved and the grade is left unchanged.
Source: Lee 2024, Food and Chemical Toxicology Industry-funded Link to this fact
The industry / Lab study (cells or chemistry)
The non-animal tests now used to set fragrance safety levels are typically off by about 2 to 3 times when predicting how potent an ingredient is.
Moderate evidence.
Caveats
Median absolute fold-misprediction 1.8-3.4; errors tend to be conservative. Natsch 2023 (same regression framework, non-fragrance-specific dataset) finds the underlying models are robust to retraining against human rather than LLNA data (only 6/139 chemicals shift >3.3-fold) — weak indirect support that this fold-error range is a stable property of the method, not an LLNA-fitting artifact, though the 62-fragrance-ingredient set itself is not retested against human data.
Source: Lee 2024, Food and Chemical Toxicology Industry-funded Link to this fact
The industry / Lab study (cells or chemistry)
Those models underestimated allergy potency by more than fivefold for 18% of the 188 chemicals tested, and more than tenfold for 11%.
Moderate evidence.
Caveats
Prediction error, not a safety-limit error; worst outliers are non-fragrance chemicals. Natsch 2023 (same author, human-anchored retraining) shows this outlier structure is not simply an artifact of fitting to mouse data: only 6/139 chemicals (4.3%) shift by more than 3.3-fold when the same models are retrained against human-anchored potency values instead of LLNA.
Source: Natsch & Gerberick 2022, ALTEX Industry-funded Link to this fact
The industry / Animal study
Older synthetic musks (musk ambrette, AETT) caused nerve damage in animal studies and were later restricted.
Moderate evidence.
Caveats
Both primaries now read directly (2026-09-26): Spencer 1979 (AETT, N=16, single lab, single route) and Spencer 1984 (musk ambrette, N=220, dual route, dose-ranging). The musk-ambrette evidence alone is large and well-controlled enough to support 'strong', but the claim bundles it with the smaller AETT study, which keeps the combined claim at moderate under the grading rule.
Source: Pinkas 2017, Environmental Research Link to this fact
The industry / Journalism
A perfumer writing in a chemistry journal said a natural ingredient such as rose oil is sometimes added in a "negligible quantity only to be able to say that you have rose oil in your formula", which "will please the marketing people" but does nothing for the scent.
Moderate evidence: Opinion; she advises against the practice and gives no frequency.
Source: Ellena 2008, Chemistry & Biodiversity Industry-funded Link to this fact
The industry / Animal study
Citronellol is treated as safe by default – GRAS at the US FDA, a 'non-toxic' biopesticide at the US EPA, and limited only by the fragrance industry's own voluntary IFRA standard – yet this government-funded study found it is brain-penetrant and calls for its content in consumer products to be regulated.
Moderate evidence: The call for regulation is the authors' conclusion from high-dose animal data; funded by the Korean Ministry of Environment, no industry money.
Source: Kim 2024, Journal of Hazardous Materials Link to this fact
The industry / Review of other studies
The industry's permitted levels for rose ketones (up to 0.07-0.8% depending on product type) overlap the concentrations at which already-sensitized volunteers reacted (0.05-0.1%), because the risk assessment is designed to prevent new allergies rather than to protect people who already have one.
Moderate evidence.
Caveats
The loosest categories are rinse-off/incidental contact where dose per area is far lower than a patch test. The design intent (induction only, not elicitation) is now confirmed by the framework's own current methodology paper (Api 2020, QRA2 §2.2); the numeric overlap still rests on the single 2007 dossier.
Source: Lalko 2007 (RIFM), Food and Chemical Toxicology Industry-funded Link to this fact
The industry / Review of other studies
The fragrance industry's safety assessment of 17 salicylates counted exposure from 10 personal-care product types only. Household cleaners and detergents, which the same report lists as uses, and breathing fragrance in were not included.
Moderate evidence.
Caveats
As of 2024, RIFM/Creme Global's aggregate exposure model (Lee 2024) does now cover household care, air care, and inhalation industry-wide, but that 2024 paper does not report salicylate-specific numbers, so it is unconfirmed whether this specific 2007 dossier's gap has since been closed for these 17 materials. A decade after this 2007 dossier, RIFM's own broader exposure model (Comiskey et al. 2017, Phase 2) had still not reached household cleaning or laundry products either — explicitly named as future work — confirming the product-type gap this claim describes persisted within the industry's own more advanced modeling program well past 2007, closing only with Lee 2024 industry-wide (still not salicylate-specific).
Source: RIFM Expert Panel / Belsito 2007, Food and Chemical Toxicology Industry-funded Link to this fact
The industry / Review of other studies
The industry panel judged each salicylate's safety one chemical at a time, although the same report says all 17 are broken down in the body to the same substance, salicylic acid.
Moderate evidence.
Caveats
The combined-exposure figure (about 1.05 mg/kg/day, group margin roughly 48-69 at the panel's own 100%-absorption default) is a evidence base calculation on the paper's Table 1, not a published number, and sums high-end maxima.
Source: RIFM Expert Panel / Belsito 2007, Food and Chemical Toxicology Industry-funded Link to this fact
The industry / Measured in people or real products
In a French fragrance-and-flavor factory, the company's own measurements found the butter-flavor chemical diacetyl (linked to 'popcorn lung' in US factory workers) in workers' breathing air at 0.05–9 ppm during short tasks; 29 of 31 samples were above the 15-minute limit of 0.1 ppm recommended by the EU's scientific committee on occupational limits.
Moderate evidence.
Caveats
Limit comparison is our (paper makes none); samplers on clothing, respirator use per task not reported; flavor chemical, factory workers, not consumers; industry-funded (V. Mane Fils).
Source: Angelini 2016, PLoS ONE Industry-funded Link to this fact
The industry / Measured in people or real products
The standard EU (REACH) screening tool for estimating workers' chemical exposure predicted lower levels than were actually measured in 61% of 431 measurements in a fragrance-and-flavor factory, sometimes by more than 100-fold.
Moderate evidence: One plant; evidence base count from the paper's S4 table (paper: 37% of tool predictions were above the measurement); short task samples.
Source: Angelini 2016, PLoS ONE Industry-funded Link to this fact
The industry / Self-reported survey
More than a third of Germans (37.8%) think people who wear perfume are more hygiene-conscious; fragrance can mask a smell but does not clean.
Moderate evidence: Attitude item; the 'does not clean' half is the author's argument, not a measurement.
Source: Klaschka 2020, Environmental Sciences Europe Link to this fact
The industry / Animal study
This lavender study was funded by a university with no declared industry involvement, unlike much of the evidence used to argue these oils are safe for children, which comes from the essential-oil trade or from a manufacturer of lavender's main constituents.
Moderate evidence.
Caveats
Funding source is not evidence quality: this study is small (n = 5 per group) and the industry studies it contrasts with are larger and guideline-compliant, though narrower in scope.
Source: Kim & Lim 2022, Journal of Korean Medical Science Link to this fact
The industry / Animal study
The industry's guideline testing cleared the two lavender molecules that were NOT the most active ones: in the study it was answering, alpha-terpineol and terpinen-4-ol were the strongest on both the estrogen and androgen receptors, and BASF tested neither.
Moderate evidence: Comparison across two papers (Hareng 2024 vs Ramsey 2019) using different assay systems; the point is the untested molecules, not a disagreement about the tested ones.
Source: Hareng 2024, Archives of Toxicology Industry-funded Link to this fact
The industry / Lab study (cells or chemistry)
The hormone safety testing on lavender and tea tree oil constituents is incomplete: no regulatory steroidogenesis assay (OECD Test Guideline 456) has been published for any of the eight internationally defined constituents of either oil, including the two that an industry laboratory reported as clear on hormone receptors.
Moderate evidence: An absence-of-evidence claim about the published record as of 2026-09; states a testing gap, not a harm. Say 'not tested', never 'found unsafe'.
The industry / Review of other studies
Fragrance safety is self-regulated: IFRA, which represents about 90% of fragrance suppliers, issues voluntary guidelines, and much of the safety research it relies on is done by the industry's own institute (RIFM) and is unpublished.
Moderate evidence: Secondary figures; the unpublished-studies point is better sourced from RIFM dossiers themselves.
Source: BCPP 2018 Link to this fact
The industry / Industry source
The industry's standard safety package for a fragrance ingredient in 1994 covered acute poisoning and skin effects (irritation, allergy, sun reactions); breathing it in was not tested, because the industry judged inhalation 'not a route of toxicological concern'.
Moderate evidence.
Caveats
As of 2024, RIFM/Creme Global's aggregate exposure model (Lee 2024) does now include a full inhalation-TTC branch across ~3,000 fragrance ingredients (99% below the inhalation TTC), so the industry itself has since closed this specific 1994 scope gap methodologically — though whether ingredients assessed only under the older dermal-only dossiers were individually revisited is not stated. The gap's closure was gradual, not a single 2024 event: Comiskey et al. 2017 (Phase 2 of the same RIFM/Creme model) added the model's first-ever inhalation branch, 23 years after this 1994 dismissal, but only for air-care and personal-care spray products — household cleaning and laundry products were explicitly still excluded and named as future work in 2017, closing only in the 2024 paper.
Source: Ford 1994 (RIFM) Industry-funded Link to this fact
The industry / Review of other studies
EPA's reviewers warned that replacing one phthalate with a close relative (DINP for DEHP, DIBP for DBP) should not be assumed to be safer.
Moderate evidence: Based on weaker data for the substitutes (fewer studies, lower exposures), not on evidence they are equally harmful.
Source: Radke 2018 (US EPA), Environment International Link to this fact
The industry / Measured in people or real products
The essential-oil industry's published defense of lavender oil in a child's breast-growth case rested on showing the child's cologne was synthetic fragrance containing a phthalate: that clears lavender oil, not the cologne.
Moderate evidence: Describes the letter's argument; one case, no dose, no biomarker; does not show the phthalate or anything else caused the breast growth.
Source: Giroux & Orjubin 2020 (JCEM letter; Consortium Huiles Essentielles), The Journal of Clinical Endocrinology & Metabolism Industry-funded Link to this fact
The industry / Lab study (cells or chemistry)
A 2026 lab test that adds simulated human liver metabolism to a DNA-damage (Ames mutagenicity) test found MORE mutagenic activity in skin creams and lotions than the conventional animal (rat) version of the same test — the opposite of metabolism detoxifying the product. Standard safety testing, including industry fragrance-material dossiers, defaults to the rat version.
Moderate evidence.
Caveats
One paper, 11 unnamed cosmetic/skin-care-cream extracts, no compound identified; not tested on perfumes or on named fragrance materials; species differences in metabolism can go either direction depending on the chemical.
Source: Schmidtmann 2026, Analytical Chemistry Link to this fact
The industry / Review of other studies
The cosmetics industry's own safety panel found that 5% benzyl alcohol and 2% benzoic acid provoke a non-allergic skin reaction (redness, swelling, itching) in most people tested with no prior sensitization needed, yet concluded both are "safe" up to 5% in cosmetics because it judged typical exposure limited, not because the reaction is rare.
Moderate evidence.
Caveats
The exposure-limited reasoning was built around occasional hair-dye use; the same industry data table lists benzyl alcohol and benzoic acid in daily leave-on products (moisturizers, body/hand skin care) without a separate exposure justification for that pattern.
Source: Nair 2001 (CIR Expert Panel), International Journal of Toxicology Industry-funded Link to this fact
The industry / Lab study (cells or chemistry)
The replacement tests are least accurate for the strongest allergens: they were wrong by about 6-fold for strong sensitizers and about 18-fold for the most extreme ones, almost always guessing them weaker than they are.
Moderate evidence: Geometric-mean fold error by LLNA potency class (Table 4); no human-outcome validation for this specific breakdown.
Source: Natsch 2015, Toxicological Sciences Industry-funded Link to this fact
The industry / Lab study (cells or chemistry)
When the same lab-dish tests that replace animal testing are checked against real human allergy-test panels instead of mouse data, they explain only about 32-45% of how strongly a chemical causes allergy in people, and the mouse test itself does barely better, explaining only 43%.
Moderate evidence: The strictly-human comparison dataset (n=62) is small and the author flags it as biased/non-representative; a larger human dataset might do better.
Source: Natsch 2023, ALTEX Industry-funded Link to this fact
The industry / Lab study (cells or chemistry)
Neither the animal test nor its lab-dish replacements are strong predictors of real human allergy thresholds; they are about equally weak, shown independently in 2015 and again in 2023.
Moderate evidence: Both comparisons reuse overlapping historical human patch-test literature (HRIPT/HMT), not independent new human data.
Source: Natsch 2023, ALTEX Industry-funded Link to this fact
The industry / Lab study (cells or chemistry)
Whether a chemical is irritating to cells in a dish, not just whether it reacts with proteins, drives a lot of the mouse allergy test's readout in ways that don't reflect real human allergy.
Moderate evidence: Statistical inference from regression coefficients across three chemical subsets, not a direct mechanistic experiment.
Source: Natsch 2023, ALTEX Industry-funded Link to this fact
The industry / Review of other studies
AETT was in fragrance use for 22 years before its neurotoxicity was documented, and cosmetic color additives are the only ingredient class the FDA actually regulates.
Moderate evidence: Regulatory-gap claim is accurate for the 1979 US framework described in the paper; verify current FDA cosmetic regulation status before using as a present-tense claim.
Source: Spencer 1979, Science Industry-funded Link to this fact
The industry / Animal study
The study's own authors, including two Research Institute for Fragrance Materials scientists, stated that the doses causing clear nerve damage in rats were 50 to 250 times their own estimate of maximum human fragrance exposure to musk ambrette, and that the lower-dose, longer-term study needed to assess real human risk was never done.
Moderate evidence: This is the study authors' own stated limitation, not an independent follow-up; no lifetime low-dose study has been conducted since (as far as this evidence base has found).
Source: Spencer 1984, Toxicology and Applied Pharmacology Industry-funded Link to this fact
Information, not medical advice. See also: myths we won’t tell you.