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.
Skin allergy / Lab study (cells or chemistry)
In the fragrance industry's own reference set of 62 fragrance ingredients, 19 are rated moderate or strong skin sensitizers, meaning chemicals capable of causing allergy.
Moderate evidence: Industry (RIFM) weight-of-evidence categories; lab and animal data, not human exposure.
Source: Lee 2024, Food and Chemical Toxicology Industry-funded Link to this fact
Skin allergy / Measured in people or real products
Allergy to these two scent chemicals rose significantly over the eight years studied (2013-2020).
Moderate evidence: Patch-test chamber system changed in 2018, which may affect the trend.
Skin allergy / Review of other studies
Benzyl salicylate, one of the most widely used fragrance chemicals (about 2,500 metric tons a year worldwide in 2007), caused skin allergy in the fragrance industry's own tests, and the industry's standards body IFRA restricts how much may be used.
Moderate evidence: Human maximization 3/50 positive at 20%; repeat-insult patch tests were negative (0/101 at 15%); LLNA EC3 1.5-2.9% in mice. Industry-funded review of largely unpublished reports.
Source: RIFM Expert Panel / Belsito 2007, Food and Chemical Toxicology Industry-funded Link to this fact
Skin allergy / Measured in people or real products
Standard allergy screening misses a large share of fragrance allergy: in one UK clinic 42% of patients allergic to at least one of the 26 EU-labeled fragrance chemicals tested negative on the standard fragrance markers, and about 4 in 10 people who reacted to their own perfume or deodorant were negative on the screen.
Moderate evidence.
Caveats
Comparator-dependent: Sukakul 2022 found ~10% missed when comparing fragrance mixes with their own ingredients (abstract-sourced). The Mann 2014 primary, now read in full (not just its abstract), confirms 203 patients reacted to any of the 26 EU fragrances and only 117 (57.6%) also reacted to a baseline marker = 42.4% missed; Mann's own Discussion states this as "40.6%" in one sentence, a small internal inconsistency in the primary (Results math gives 42.4%). More importantly, Mann's own paper reports a similarly-designed Danish study (Heisterberg 2011, n=1,508, also routine testing of all patients with the same 26 fragrances) that found only 15.6% missed — a ~2.5-fold discrepancy the authors attribute only partly to higher test concentrations and a different patient population, not fully resolved. The own-product figure (41%/45%) is from a different study (Uter 2007), not Mann.
Skin allergy / Measured in people or real products
Some fragrance chemicals now banned from EU cosmetics sensitized almost every volunteer in the industry's own tests: diethyl maleate 25 of 25, dihydrocoumarin 25 of 25, alantroot oil 23 of 25.
Moderate evidence: Maximization test is deliberately exaggerated (irritant pre-treatment, occlusion, ~10x use level); substances are banned and mostly left perfumery before 2000.
Skin allergy / Measured in people or real products
Essential oils are made largely of the same chemicals labeled as fragrance allergens: supplier data sheets give sweet-orange oil about 95% limonene, lavender oil about 40% linalool, and palmarosa oil up to 84% geraniol.
Moderate evidence: Composition from supplier safety data sheets, which vary between suppliers (cypress oil limonene 5-7% vs 14%); not measured in finished products.
Source: Klaschka 2016, Environmental Sciences Europe Link to this fact
Skin allergy / Lab study (cells or chemistry)
In the industry's own analysis, chemicals that only become allergens after they are changed by air or by the skin's own enzymes were the worst-predicted group (4.5- to 6.5-fold error, versus about 2-fold for directly reactive chemicals); the dish assays have little capacity to oxidize or metabolize chemicals the way air and skin do.
Moderate evidence: N=32 chemicals in the worst domain; TIMES-software-assigned mechanistic class, not exhaustively confirmed in vivo.
Source: Natsch 2015, Toxicological Sciences Industry-funded Link to this fact
Skin allergy / Lab study (cells or chemistry)
Oxidized linalool and oxidized limonene, among the most common fragrance allergens found in European skin clinics, are not in the 312-chemical dataset these safety models were built on.
Moderate evidence: Clinic prevalence figures for oxidized terpenes come from a separate paper (Sukakul 2022), not this one.
Source: Natsch 2015, Toxicological Sciences Industry-funded Link to this fact
Skin allergy / Measured in people or real products
For two common fragrance chemicals, trans-2-hexenal and methyl-2-nonanoate, the mouse test underestimated true human skin-sensitizing potency by 26- to 42-fold — the error direction that could make an industry safety limit too permissive rather than too strict.
Moderate evidence.
Caveats
Only 2 of 57 chemicals in the dataset show this specific magnitude of underestimation; industry-funded (RIFM/DABMEB); the human NOEL itself is a floor, not a measured breakpoint, since dosing was not escalated to a positive result for ethical reasons.
Source: Api, Basketter & Lalko 2014, Cutaneous and Ocular Toxicology Industry-funded Link to this fact
Skin allergy / Measured in people or real products
A Givaudan-funded study that chemically measured 39 real aged fragrance products found linalool hydroperoxide levels far below the dose shown to trigger a reaction in already-allergic patients, roughly 180 times lower on average.
Moderate evidence: Industry-funded (Givaudan), single study, not independently replicated; only two product categories tested (hydroalcoholic fine fragrance, antiperspirant/deodorant).
Source: Kern 2014, Analytical and Bioanalytical Chemistry Industry-funded Link to this fact
Skin allergy / Measured in people or real products
In a UK study that routinely patch tested 1,951 eczema patients with both the standard fragrance-allergy screen and all 26 individually EU-labeled fragrance chemicals, 42% of the patients allergic to at least one of those 26 chemicals would have been missed if only the standard screen had been used.
Moderate evidence.
Caveats
Single UK tertiary-referral clinic (n=1,951), not a general-population figure. A comparably-designed Danish study using nearly the same routine-testing protocol found only 15.6% missed, so this exact percentage does not generalize across clinics without qualification.
Skin allergy / Measured in people or real products
A large fragrance-industry study tested 104 real consumer products, including creams, lotions, and antiperspirants, for the allergenic chemicals that form when the scent chemicals limonene and linalool oxidize in air, and found almost none of them: 96% of products had no detectable hydroperoxide, and the one confirmed positive was hundreds of times below the dose known to trigger a reaction.
Moderate evidence: Industry-funded (IDEA consortium: Givaudan, Firmenich, IFF, IFRA), not independently replicated by a non-industry group.
Source: Natsch 2019, Food and Chemical Toxicology Industry-funded Link to this fact
Skin allergy / Measured in people or real products
Even when scientists chemically tested the actual products that people with a confirmed allergy to oxidized limonene or linalool blamed for their skin reaction, none of the 28 products tested (11 from confirmed allergy patients) contained a detectable amount of the allergenic chemical.
Moderate evidence.
Caveats
Small sample (28 products, 11 from confirmed patch-test-positive patients) relative to the >22,000-patient clinical literature on this allergy; industry-funded, not independently replicated; other exposure routes (skin-surface formation after application, airborne deposition) are untested.
Source: Natsch 2019, Food and Chemical Toxicology Industry-funded Link to this fact
Skin allergy / Measured in people or real products
Fragrance allergy caused by a deodorant correlated with a positive patch test to fragrance mix II or HICC (34.3% and 24.9% of deodorant-linked cases) more than to fragrance mix I or Myroxylon pereirae (28.2% and 12.7%), while scented lotion and shampoo correlated more with fragrance mix I and M. pereirae, suggesting deodorants are enriched for the specific chemicals in fragrance mix II rather than just containing more fragrance overall.
Moderate evidence: Correlational, single retrospective dataset; the paper does not independently verify deodorant formulations chemically, relying on prior surveys for that.
Source: Heisterberg 2011, Contact Dermatitis Link to this fact
Skin allergy / Review of other studies
Reviewing all available human, animal, and lab-dish evidence on 106 well-studied fragrance chemicals, the fragrance industry's own research institute found that 90 of them (85%) are capable of causing allergic skin sensitization at some dose — 6 strong, 23 moderate, 41 weak, and 26 very weak sensitizers — and only 10 had no evidence of sensitization at all.
Moderate evidence.
Caveats
Industry-authored (RIFM/Givaudan-consultant); the 106 materials were chosen specifically because in vivo data already existed on them, so this is not a random or representative sample of fragrance ingredients in commercial use, and none was rated the most severe ("extreme") category.
Source: Na et al. 2022, Dermatitis Industry-funded Link to this fact
Skin allergy / Measured in people or real products
In this industry dataset, the fragrance chemical methyl-2-nonynoate was rated a strong skin sensitizer with a human no-effect dose of only 24 micrograms per square centimeter of skin — among the lowest doses in the whole 106-chemical set — while a chemical like ethylene brassylate needed over 23,000 micrograms per square centimeter, nearly a thousandfold more, before it showed no allergic reaction.
Moderate evidence: Single industry dataset (RIFM Database plus published literature); NOEL is the maximum dose tested without inducing sensitization, not necessarily the true threshold.
Source: Na et al. 2022, Dermatitis Industry-funded Link to this fact
Skin allergy / Lab study (cells or chemistry)
For several fragrance chemicals with no allergy-triggering chemical structure in their original (parent) form — including linalool and D-limonene, two of the most widely used fragrance ingredients in the world — the fragrance industry's own computer models predict that their breakdown products (metabolites) can trigger skin allergy, even though the parent chemicals were classified as non-sensitizers overall.
Moderate evidence.
Caveats
This paper's human no-effect-level data for linalool and limonene do not specify the oxidation state of the test material used; independent clinical studies already in this evidence base's contradiction register find that once these two chemicals are exposed to air, their oxidation products are among the commonest fragrance contact allergens in European patch-test clinics.
Source: Na et al. 2022, Dermatitis Industry-funded Link to this fact
Skin allergy / Other evidence
Of 226 chemicals identified in 71 deodorants, 84 (37%) carried at least one chemical-structure 'alert' for skin-protein reactivity, and 70 of those were aldehydes, ketones, or alpha,beta-unsaturated aldehydes/ketones/esters – the reactive chemical classes already established as the dominant fragrance-allergen chemotype.
Moderate evidence.
Caveats
A structural alert is a hazard flag based on molecular structure only; the paper explicitly states it gives no information on potency (strong/moderate/weak) and is not a substitute for patch testing or animal studies.
Source: Rastogi et al. 1998 (Contact Dermatitis 39:293), Contact Dermatitis Link to this fact
Skin allergy / Other evidence
Citral, a fragrance chemical already established as a human skin sensitizer, was detected by GC-MS in 26 of 71 European deodorants tested in 1998 (15 of 22 vapo-sprays, 8 of 22 aerosol sprays, 3 of 27 roll-ons).
Moderate evidence: Presence only, not concentration; a single 1998 market snapshot from 5 European countries.
Source: Rastogi et al. 1998 (Contact Dermatitis 39:293), Contact Dermatitis Link to this fact
Skin allergy / Measured in people or real products
In a 1996 chemical analysis of 73 European deodorants, the seven allergens used in the standard fragrance-mix patch test were very unevenly present: geraniol was found in 76% of products and eugenol in 57%, but cinnamic aldehyde in only 17% and isoeugenol in 29%.
Moderate evidence.
Caveats
Single study, one product category, purchased 1996 in 5 European countries; not replicated for deodorants specifically, though the pattern matches the same lab's later household-product survey (Rastogi 2001).
Source: Rastogi et al. 1998, Contact Dermatitis Link to this fact
Skin allergy / Measured in people or real products
In a controlled study, deodorants containing hydroxycitronellal at 0.032% to 0.32% triggered allergic armpit rashes within a few weeks in all 7 people already allergic to the chemical, while an identical unscented deodorant on the other armpit, and the same deodorants used by 7 non-allergic people, caused no reactions (P<0.001).
Moderate evidence.
Caveats
Single-center study (Malmö only), unlike the 5-center Bruze 2003 cinnamic-aldehyde companion study graded strong; small cohort (7 test + 7 control patients), though statistically significant with a concurrent negative control group and blinded, randomized contralateral-axilla design.
Source: Svedman et al. 2003, Contact Dermatitis Link to this fact
Skin allergy / Other evidence
162 individual fragrance chemicals have documented reports of causing contact allergy or allergic contact dermatitis, but routine patch testing screens for only a handful of markers (fragrance mix I, fragrance mix II, balsam of Peru, HICC, and the EU's 26 labeled allergens).
Moderate evidence: A single author's own literature tally, not a systematic review with stated search criteria; the same paper's abstract rounds this down to "more than 150".
Source: de Groot 2020, Dermatitis 31(1):13-35, Dermatitis® Link to this fact
Skin allergy / Measured in people or real products
Patients allergic only to oxidized linalool also tended to react to a wide range of other cosmetic ingredients (preservatives such as parabens and MCI/MI, colophony, and most fragrance-mix ingredients), a pattern the researchers link to using multiple scented or cosmetic products rather than the chemicals being related to each other.
Moderate evidence: Patch-test associations, not confirmed shared product exposures; the authors' own interpretation, not chemically verified.
Source: Sukakul 2024, Contact Dermatitis Industry-funded Link to this fact
Skin allergy / Lab study (cells or chemistry)
At least six substances on the EU's official list of 26 named fragrance allergens (geraniol, cinnamaldehyde, eugenol, isoeugenol, cinnamyl alcohol, and anisyl alcohol) were included among the 56 chemicals tested for standardized 24-hour human-skin penetration in this 2020 industry dataset, but the paper's main text and figures report only their relative ranking (not exact absorption percentages) among the 56 chemicals; the individual percentages exist only in a supplementary table not available in this ingest.
Moderate evidence.
Caveats
Exact per-chemical dermal-delivery percentages for these fragrance ingredients were not accessible in the extracted text; only their approximate rank position among the 56 tested chemicals is reported here.
Source: Hewitt 2020, Journal of Applied Toxicology Industry-funded Link to this fact
Skin allergy / Lab study (cells or chemistry)
Three chemicals (propylparaben, geraniol, and benzophenone) penetrated human skin markedly less when dissolved in ethanol than when dissolved in a water-based buffer (PBS) in this standardized 24-hour skin-penetration study, apparently because ethanol evaporates faster and can let the chemical precipitate on the skin surface before it is absorbed; a fourth chemical, hydrocortisone, showed no such difference between the two solvents.
Moderate evidence.
Caveats
Single industry-funded lab; n=12 replicates per condition (3 skin discs x 4 donors); the exact percentage difference for geraniol specifically is not given as a number in the accessible text, only described qualitatively as lower in ethanol.
Source: Hewitt 2020, Journal of Applied Toxicology Industry-funded Link to this fact
Skin allergy / Lab study (cells or chemistry)
In lab tests on human skin, how much of the lavender-scented chemical linalool actually soaked in ranged from under 2% to nearly 15% of the amount applied, depending only on what solvent it was dissolved in and whether the skin was covered.
Moderate evidence: Single in-vitro study, one donor pool, fresh (non-oxidized) linalool at a fixed 4% test concentration in laboratory solvents, not any real product formula.
Source: Brain et al. 2022 (An-eX/RIFM), International Journal of Pharmaceutics Industry-funded Link to this fact
Skin allergy / Other evidence
Cutting freshly dyed hair bare-handed leaves hair-dye chemicals on hairdressers' hands: 23 of 29 hairdressers had measurable dye on their hands after cutting, and 11 of 12 samples of newly dyed hair cuttings contained dye compounds even after rinsing and shampooing.
Moderate evidence.
Caveats
The after-cutting sample covers everything since application (rinsing, shampooing, combing), not cutting alone; the serving-hand difference was significant only for resorcinol (P<0.05).
Source: Lind et al. 2005, The Annals of Occupational Hygiene Link to this fact
Skin allergy / Other evidence
Hairdressers' hands carried measurable hair-dye chemicals even before they started mixing dye (PPD in 3 of 33, 3-aminophenol in about 20 of 33), pointing to carryover from earlier clients and contaminated trolleys, dryers, and phones.
Moderate evidence: Nine hairdressers had dyed earlier the same day; surface contamination data were not shown; background and task exposure can't be fully separated.
Source: Lind et al. 2005, The Annals of Occupational Hygiene Link to this fact
Skin allergy / Other evidence
Dyes used on polyester, acetate, and nylon clothing (disperse dyes) are not chemically bound to the fiber, so they can rub or wash onto skin; they are not used on cotton, wool, or linen.
Moderate evidence: Review statement citing industry and textbook sources; says nothing about how much transfers in real wear.
Source: Malinauskiene 2013, Contact Dermatitis Link to this fact
Skin allergy / Other evidence
Reported clinical relevance of disperse dye patch-test reactions was 75 to 100% in aimed studies and 42 to 78% in screening studies, but it was mostly judged by the rash clearing after dark synthetic clothing was stopped, and chemical analysis of garments rarely found the dye the patient reacted to.
Moderate evidence: Authors say relevance may be overestimated; garment analyses are few (latest 2000, plus Malmo 3 of 121 items), so neither side is well measured.
Source: Malinauskiene 2013, Contact Dermatitis Link to this fact
Skin allergy / Measured in people or real products
In one tertiary referral clinic's 821 consecutive dermatitis patients, 9.4% reacted to limonene hydroperoxides and 11.7% to linalool hydroperoxides, the allergens that form when these scent chemicals react with air; most reactions were weak.
Moderate evidence.
Caveats
Single tertiary referral center, so higher than the 2.5-5.4% and 3.9-7.7% the authors cite from other studies; the patients were already suspected of contact allergy, not the general public; a further 17% and 22% of readings were doubtful. Test preparations were bought from Chemotechnique and the authors declare no conflicts.
Source: Dittmar & Schuttelaar 2019, Contact Dermatitis Link to this fact
Skin allergy / Measured in people or real products
In a Danish university clinic, 15.7% of 6,004 consecutive dermatitis patients (2010-2015) were allergic to at least one of the 26 fragrance chemicals that EU cosmetics must name on the label.
Moderate evidence: Single clinic of patients referred for dermatitis, not the general population; 25 of the 26 allergens tested.
Skin allergy / Other evidence
The main US dermatology society's extended allergy-test panel now uses air-oxidized limonene (0.2%) and linalool (0.5%) rather than the fresh chemicals, because testing the oxidized form finds more relevant reactions.
Moderate evidence: Committee consensus relaying two de Groot 2019 reviews, not new data; the US doses are lower than the European EECDRG doses (1% and 0.3%) because of irritant-reaction risk.
Skin allergy / Other evidence
About 1 in 10 people referred to North American skin-allergy clinics and patch tested in 2021-2022 (308 of 3,040, 10.1%) reacted to hydroperoxides of linalool, the air-oxidized form of a very common scent chemical; 5.7% (173 of 3,047) reacted to oxidized limonene.
Moderate evidence.
Caveats
Referral clinic patients, not the general public; most reactions were weak (+) or doubtful, and the authors say it is unclear whether these reflect true allergy or irritation (irritant reactions 2.0% and 0.9%). Single network, one two-year cycle.
Skin allergy / Other evidence
Wet workers reacted to medium-potency allergens about 1.3 times as often as dry workers (25.3% vs 20.8%; odds ratio 1.29, 95% CI 1.01-1.66), and to the rubber chemical thiuram mix 2.8 times as often (8.7% vs 3.0%).
Moderate evidence: Borderline interval; occupation assigned by one investigator from current job only; glove rubber exposure explains part of the result.
Source: Kohli & Nedorost 2016, Journal of the American Academy of Dermatology Link to this fact
Information, not medical advice. See also: myths we won’t tell you.