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.
Hidden ingredients / Measured in people or real products
Styrene migrates from polystyrene food-contact packaging into food; US dietary intake from this route has been estimated at roughly 9 micrograms per day (Lickly et al. 1995), with strawberries and avocados showing the highest detected styrene concentrations in an FDA Total Diet Study of 70 foods.
Moderate evidence.
Caveats
Modeling-based estimate combining migration data and consumption factors, not a direct measurement of total dietary intake in a real population; food styrene can also arise from natural/other sources, not only packaging migration.
Source: IARC 2019 (Vol. 121) Link to this fact
Hidden ingredients / Government agency
EPA lists moth repellents and air fresheners among household-product sources of organic-gas (VOC) indoor air pollution, and notes that paradichlorobenzene, a chemical carrying an 'avoid breathing vapors' warning on moth-repellent labels, is also the key active ingredient in many air fresheners.
Moderate evidence: EPA names the shared ingredient and the moth-repellent warning; it does not separately assess air-freshener-specific health risk.
Source: US EPA, "The Inside Story" Link to this fact
Hidden ingredients / Measured in people or real products
In a database analysis of Korean household chemical products, the most prevalent chemicals flagged as endocrine disruptors were compounds added as fragrance ingredients or preservatives (e.g. linalool, D-limonene, hexyl cinnamaldehyde), not the product's main functional ingredient.
Moderate evidence.
Caveats
Ingredient-list matching against EPA CompTox/ToxCast hazard flags, not measured concentrations or confirmed toxicity in the products as sold; single-country (South Korea) database of household chemical products (air fresheners, cleaners, detergents, fabric softeners, deodorants), not personal-care/cosmetic products specifically.
Source: Lee & Ji 2022, Ecotoxicology and Environmental Safety Link to this fact
Hidden ingredients / Measured in people or real products
Diethyl phthalate (DEP), the phthalate most often found in fragrance products, most likely comes from the fragrance ingredient itself, where it functions as a solvent and fixative rather than from another use in the formulation.
Moderate evidence.
Caveats
This is the authors' interpretive conclusion drawn from DEP's concentration pattern across product categories (highest in fragrance products, by far), not a direct mechanistic test of fragrance-longevity. The 'fixative' function (slowing evaporation of more volatile scent components, making scent last longer) is standard perfumery-formulation terminology, consistent across sources, but this specific single 2006 FDA study is the only one in our evidence base that states the fixative role explicitly; the fragrance-concentration finding itself is independently replicated by two other surveys (Koniecki 2011, Dodson 2012) already in our evidence base.
Source: Hubinger & Havery 2006, Journal of Cosmetic Science Link to this fact
Hidden ingredients / Review of other studies
An FDA laboratory survey concluded that phthalate esters found in most cosmetic products most likely enter via the fragrance ingredient itself, where they function as a solvent/fixative rather than for some other purpose.
Moderate evidence.
Caveats
This is the study authors' own interpretive conclusion (source attribution), not a direct measurement of fixative function or of scent longevity; it is consistent with, but does not by itself prove, the specific mechanism that phthalates make fragrance 'last longer'. Based on a single FDA survey (n=48 products, 2006), though its DEP findings are independently corroborated by later, larger surveys (Koniecki 2011, n=252).
Source: Hubinger & Havery 2006, Journal of Cosmetic Science Link to this fact
Hidden ingredients / Measured in people or real products
This industry-funded skin-irritation study of 24 detergent formulations reports no fragrance or enzyme content for any tested product; formulation details, including whether products were fragranced or enzyme-containing, are described only as 'available upon request' from the funding trade association, not disclosed in the published paper.
Moderate evidence: Absence of disclosure in a single paper; cannot confirm whether fragrance/enzyme content varied across the 24 formulations or influenced the irritation results.
Source: Robinson et al. 2005, Food and Chemical Toxicology Industry-funded Link to this fact
Hidden ingredients / Measured in people or real products
Electron microscopy found nanosized particles or nanostructures in two spray products that were NOT labeled or marketed as nanotechnology-based (a regular silver spray and a regular disinfectant spray), while three products that WERE labeled nanotechnology-based (facial, hair, and skin-hydrating sprays) showed no detectable particles or structures under the same microscopy.
Moderate evidence: Morphological/TEM evidence only, not chemical confirmation of nanomaterial identity; single-lab study of 10 products.
Source: Nazarenko 2014, Environmental Science: Nano Link to this fact
Hidden ingredients / Measured in people or real products
In a 2007 NRDC lab test, none of 14 US air fresheners listed phthalates as an ingredient on the label, and 10 of the 14 listed no ingredients at all.
Moderate evidence: Single lab, one sample per product, not peer reviewed or independently replicated since 2007.
Source: NRDC 2007, Clearing the Air Link to this fact
Hidden ingredients / Review of other studies
d-Limonene, a common fragrance/scent chemical, was detected in 100% of personal and area air samples in 7 nail salons despite not appearing as an ingredient on any safety data sheet the authors reviewed for the nail products used there.
Moderate evidence.
Caveats
Authors could not identify the actual source (possible fragrance classification, or an unreviewed product category such as lotions, scrubs, waxes or cleaning products); small pilot (7 salons).
Hidden ingredients / Measured in people or real products
Di(2-ethylhexyl) phthalate (DEHP), a plasticizer used to make nail polish more flexible and durable, was the most abundant of six phthalates measured on silicone wristbands in all 10 nail salons tested (median 391.80 ng/g-wristband), with diethyl phthalate (DEP) second highest (median 154.4 ng/g-wristband) — but the paper never discusses what produces its measured DEP, unlike its explicit attribution of DEHP to nail polish and d-limonene to cleaning/fragrance products.
Moderate evidence.
Caveats
Small sample (n=10 for phthalates); wristband measurement is a sorbed dose, not a direct air concentration; the paper's own figures and Discussion text mislabel this unit as ppb instead of the Methods/footnote's ng/g-wristband, an internal inconsistency in the published article.
Source: Han et al. 2022 Link to this fact
Hidden ingredients / Review of other studies
Danish product checks cited in a 2004 review found parabens in 77% of 215 cosmetics (0.01-0.87%) and 87% of skin creams, with ingredient labels wrong about parabens in about 10% of products.
Moderate evidence: Secondhand from Rastogi 1995 and 2000, and from 1990s Danish and Swedish products; not re-checked in this evidence base; current formulations differ.
Source: [[src-2004-endocrine-disrupters-estrogenic-chemicals-body-care-cosmetics|Harvey & Darbre 2004]], Journal of Applied Toxicology Link to this fact
Hidden ingredients / Other evidence
Under an FDA bulk-labeling rule the flavor industry relies on, a multi-ingredient flavor can carry one sentence ('All flavor ingredients contained in this product are approved for use in a regulation of the Food and Drug Administration') instead of a list of each ingredient, provided every flavor ingredient is approved in an FDA regulation; non-flavor ingredients and flavor ingredients outside such a regulation must be listed separately.
Moderate evidence.
Caveats
1995 description by FEMA's own authors, who add that FDA treated FEMA GRAS lists as reliable industry lists for this purpose. The 2024 CFR text of 21 CFR 101.22(g)(2) names only FDA regulations and says nothing about industry lists.
Source: Hallagan 1995, Regulatory Toxicology and Pharmacology Industry-funded Link to this fact
Hidden ingredients / Other evidence
The FEMA panel does not require toxicology tests on each flavor ingredient: for the 119 ring-structured terpene flavors it judged safety mainly from the metabolism and tests of a few parent compounds (such as menthol, carvone, camphor-type ketones) and from low estimated intake.
Moderate evidence.
Caveats
This is the industry panel's own method statement; the paper argues it is scientifically sound because related substances share metabolic pathways, and does not show case-by-case failure.
Source: Adams 1996, Food and Chemical Toxicology Industry-funded Link to this fact
Hidden ingredients / Other evidence
Several flavor chemicals in this review (menthol, carvone, camphor, borneol, terpineol, ionones, damascones and the musks muscone, and civettone) are also fragrance materials, but the FEMA review assesses them only as swallowed food ingredients; it gives non-oral data 'less weight' and has no inhalation or skin-allergy assessment.
Moderate evidence: The paper says some inhalation data may substitute when flavor exposure is far smaller than inhaled exposure; it does not claim any airborne or skin risk from these substances.
Source: Adams 1996, Food and Chemical Toxicology Industry-funded Link to this fact
Hidden ingredients / Other evidence
Most nasal products in this review's 1999 US excipient table (14 of 19) contained benzalkonium chloride, including all non-steroid products listed and 7 of 12 corticosteroid entries; the propellant inhalers and Rhinocort did not.
Moderate evidence: 1999 US market snapshot of listed brands only; formulations have changed since.
Source: Bernstein 2000, Journal of Allergy and Clinical Immunology Link to this fact
Hidden ingredients / Other evidence
Under the flavor panel's 2005 procedure, the part of an essential oil whose chemistry has not been identified is treated as the highest-concern structural class and held to 90 micrograms a day, unless people already eat far more of the plant (5-10 times) than of the added oil.
Moderate evidence: Describes the rule as written by the industry panel; the paper shows no example of an oil failing it. 90 µg/day is the Munro 1996 Class III threshold. Industry-authored (FEMA).
Source: Smith 2005, Food and Chemical Toxicology Industry-funded Link to this fact
Hidden ingredients / Other evidence
By FEMA's 2009 count there are more than 2,400 single flavoring chemicals and about 300 natural flavor complexes in US use, more than 2,700 holding FEMA GRAS status, and about 2,000 of them have no explicit FDA regulation, so the Expert Panel's conditions of use are the main rule for them.
Moderate evidence.
Caveats
Rounded industry counts from 2009; the paper's own totals are only approximately consistent. Written by two FEMA employees and supported by FEMA (the flavor industry trade association); it is the industry's own account of its own program.
Source: Hallagan 2009, Food and Chemical Toxicology Industry-funded Link to this fact
Hidden ingredients / Other evidence
Only about 3 of roughly 200 food ingredients that FDA had acknowledged through its voluntary GRAS notification program by 2009 were FEMA GRAS flavor ingredients; the authors say flavors rarely use the FDA route because a FEMA GRAS number already has commercial value.
Moderate evidence.
Caveats
Counts as of 2008; reasons given are the authors' own. Written by two FEMA employees and supported by FEMA (the flavor industry trade association); it is the industry's own account of its own program.
Source: Hallagan 2009, Food and Chemical Toxicology Industry-funded Link to this fact
Hidden ingredients / Other evidence
In 2008, about one in five (19.5%, 6,463 of 33,212) cosmetics and personal care products in the US FDA's voluntary registry contained a formaldehyde-releasing preservative, including 27.3% of rinse-off products and 16.9% of leave-on products.
Moderate evidence: The registry is voluntary, so it may not represent all US products; 2008 data, and the authors found no comparable European database.
Hidden ingredients / Other evidence
Formaldehyde-releasing preservatives are not labeled 'formaldehyde'; the commonest in US products were imidazolidinyl urea (7.0%), DMDM hydantoin (5.4%, and 12.6% of rinse-off products), diazolidinyl urea (4.5%) and quaternium-15 (1.4%).
Moderate evidence: Registry listing, not content analysis; free formaldehyde levels in products are covered in Part 2, not this paper.
Hidden ingredients / Other evidence
In a 2010 Danish store survey, 19 of 1,272 cosmetic labels (1.5%) listed MI alone; chemical analysis of the products found 14 of the 19 had more than 50 ppm and 5 had about 100 ppm, in rinse-off products, leave-on creams, and wet wipes (2 contained none detectable).
Moderate evidence: One country, one month (October 2010), 9 stores, labels possibly duplicated across stores of the same owner; not a market-wide share, and the products are not named.
Source: Lundov 2011, British Journal of Dermatology Link to this fact
Hidden ingredients / Other evidence
The workers were challenged with commercial cleaning products that contained mixtures of irritants and possible sensitizers, so the authors could not formally identify which ingredient caused each reaction.
Moderate evidence: Authors' own limitation statement; they infer QACs from known asthma-hazard predictions (asthma hazard index 0.81-0.95), which is a model score, not a test of the product.
Hidden ingredients / Other evidence
Textile-dye allergy often looks atypical (purpuric, hives-like, prurigo-like, eczema away from skin contact points) and a high index of suspicion is needed, so it can be missed.
Moderate evidence: Based on case reports and case series; no frequency of atypical presentations across all patients.
Source: Malinauskiene 2013, Contact Dermatitis Link to this fact
Hidden ingredients / Other evidence
The same audit estimated, with 95% confidence, that at least 62% of the 8,105 chemicals allowed in US food lack a published feeding study, and at least 54% of directly added additives.
Moderate evidence: Sampling estimate, not chemical-by-chemical search; flavor-industry literature reviews are under-captured by the databases, so the flavor gap may be overstated.
Source: Neltner et al. 2013, Reprod Toxicol, Reproductive Toxicology Link to this fact
Hidden ingredients / Other evidence
FDA's own database of almost 4,000 chemicals allowed directly in food had a staff comment of 'no or insufficient toxicology data' for almost 25% of them (2013 review of the database).
Moderate evidence: Counts the agency's database comments, as reported by the Pew authors; data gap, not proof of hazard. More detail in Neltner 2013 (data gaps).
Source: Maffini 2013, Comprehensive Reviews in Food Science and Food Safety Link to this fact
Hidden ingredients / Measured in people or real products
In a March 2013 label survey of 1,795 cosmetics in 14 Copenhagen supermarkets, 60 (3.3%) listed MI as a preservative, up from 1.5% in a similar survey three years earlier, and a larger share were leave-on (42% vs about 16%).
Moderate evidence: Label counts only, one city, one month; concentrations unknown; comparisons are with surveys by the same group and different methods in Sweden (0.5%).
Hidden ingredients / Animal study
In a Dutch government study, isoeugenol was the only one of five skin-allergy fragrance chemicals to stimulate the lymph nodes of mice that breathed it (about 7-fold more cell growth after 6 hours a day for 3 days), while limonene and linalool did not.
Moderate evidence.
Caveats
Mice (6 per group), one lab, nose-only at 75 mg/m3, an unvalidated assay tested mostly on strong sensitizers; a hazard signal, not human disease. Government-funded (NVWA), no industry money declared.
Source: ter Burg 2014, Inhalation Toxicology Link to this fact
Hidden ingredients / Other evidence
Dyes and finishing chemicals in clothing are not declared on labels; EU rules from May 2012 require only fiber composition and country of origin, so doctors can hardly detect new textile allergens or the disappearance of old ones.
Moderate evidence: The authors' regulatory description of Regulation (EU) 1007/2011 as of 2013-14; not a survey of labels, and later rules may differ.
Hidden ingredients / Other evidence
The paper's own authors state that its numbers for the fragrance chemical 2-phenylethanol are not an accurate picture of real exposure, because concentration data came from two companies and every product was assumed to contain it.
Moderate evidence: Statement is the authors' own caution; it cuts both ways (could over- or under-estimate).
Source: Safford et al. 2015, Regulatory Toxicology and Pharmacology Industry-funded Link to this fact
Hidden ingredients / Other evidence
The industry method accepts a natural plant-oil flavor if each chemical family's estimated daily intake stays under generic thresholds of 1,800, 540, or 90 micrograms per person per day (by structural class), and the unidentified remainder stays under 90.
Moderate evidence: Thresholds are screening values from a database of pesticide, drug and herbicide studies, not tests on the flavor; failing leads to more data, not a ban. Industry-authored (FEMA).
Source: Cohen 2018, Food and Chemical Toxicology Industry-funded Link to this fact
Hidden ingredients / Review of other studies
Severe food allergic reactions, including anaphylaxis, have been traced to minor ingredients that need not be on an allergen label, such as mustard, celery (in some countries), lupin, fenugreek, spices, and natural colorings, which is why some anaphylaxis is labeled 'idiopathic'.
Moderate evidence: Mostly case reports; the review gives no overall rate of hidden-allergen anaphylaxis and has no stated search method or appraisal.
Source: Skypala 2019, Frontiers in Immunology Link to this fact
Hidden ingredients / Other evidence
Methylisothiazolinone allergy became an epidemic after MI alone was approved as a preservative at higher concentrations in the early 2000s, and a patch test with only the MCI/MI mix can miss it; hidden sources include moist toilet paper, makeup-remover wipes, paint, and homemade slime.
Moderate evidence: Narrative review; the 'up to 8%' MCI/MI contact-allergy figure has no stated population in the text.
Source: Militello 2020, Dermatologic Clinics Link to this fact
Hidden ingredients / Industry source
The flavor industry's own expert panel, funded by flavor and essential-oil trade groups, re-affirmed eight clove, cinnamon leaf, and West Indian bay leaf flavorings as 'generally recognized as safe' in 2020; the same oils were first declared GRAS in 1965, so the panel was re-affirming its own decision.
Moderate evidence: The industry's own account of its own GRAS determination; no new experiments or independent review.
Source: Gooderham 2020, Food and Chemical Toxicology Industry-funded Link to this fact
Hidden ingredients / Other evidence
The natural clove, cinnamon leaf, and West Indian bay leaf oils contain methyl eugenol, estragole, or safrole, compounds that cause liver cancer in rodents at high doses; the panel estimates cinnamon leaf oil's safrole intake at about 6 micrograms/person/day, around 40 times its own 0.15 microgram genotoxic threshold, and clears it on a margin-of-exposure argument (above 19,000).
Moderate evidence.
Caveats
Margins are above the 10,000 benchmark the panel cites from EFSA; the intake figures are from an industry poundage survey, and rodent carcinogenicity at high doses is not human cancer evidence.
Source: Gooderham 2020, Food and Chemical Toxicology Industry-funded Link to this fact
Hidden ingredients / Other evidence
The FEMA panel assesses roughly 3,000 individual flavor ingredients, not the finished flavor mixtures (which can hold 100 or more ingredients) that are added to foods; the authors call evaluating each mixture 'impracticable' and say there is 'no indication' an ingredient behaves differently in a mixture.
Moderate evidence: The 'no indication' statement is an assertion with no data or citation in the paper; interactions in mixtures are not tested by the program as described.
Source: Hallagan 2020, Food and Chemical Toxicology Industry-funded Link to this fact
Hidden ingredients / Other evidence
FEMA has removed at least 14 substances from its GRAS list over the years, including musk ambrette, quinoline, calamus, ethylene oxide, and methyl eugenol; the most common reason was that existing data no longer supported a conclusion of general recognition of safety, not newly shown harm.
Moderate evidence: List and reasons are FEMA's; the paper does not date each removal or give the data behind it. The count of 14 treats calamus and calamus oil as one.
Source: Hallagan 2020, Food and Chemical Toxicology Industry-funded Link to this fact
Hidden ingredients / Other evidence
BAK is an ingredient in many products people do not think of as 'preservative' sources, including eye drops, antiseptic creams and dressings, bath oils and emollients, makeup remover, hand sanitizers, disinfectant wipes, laundry rinses, and plaster of Paris.
Moderate evidence: From case sources in one clinic; presence in a product does not mean it will cause allergy in most users.
Hidden ingredients / Other evidence
Formaldehyde can be present in a product without being declared: it was found above 2 ppm in nail polishes marketed as formaldehyde-free, and it can form from releasers, surfactants, or packaging resins in cosmetics.
Moderate evidence: Review statement citing other studies; amounts in typical products are not given beyond the nail-polish figure.
Hidden ingredients / Other evidence
Oak moss absolute, jasmine absolute, tuberose oil, cassie absolute, and wintergreen oil, all perfumery materials, are also US food flavorings; the 2020 industry survey reports only tiny volumes (oak moss 34 kg, cassie 29 kg, jasmine absolute 100 kg, tuberose 0.8 kg a year, wintergreen oil 700 kg), or about 0.08 to 72 millionths of a gram per person per day.
Moderate evidence: Food-use volumes from a voluntary industry survey (corrected by an assumed 80% response); says nothing about fragrance volumes, where these materials are used far more.
Source: Guengerich 2026 (FEMA Expert Panel), Food and Chemical Toxicology Industry-funded Link to this fact
Hidden ingredients / Other evidence
The flavor panel's food assessment is silent on skin allergy, inhalation, and cosmetic use: the words allergy, sensitization, and atranol do not appear in the assessment, and the abstract excludes 'dietary supplements or any products other than food'; so food-GRAS status of oak moss or jasmine says nothing about their behavior as fragrance allergens.
Moderate evidence: Absence of a topic in an industry review is not evidence of harm; it is a scope limit the paper states for supplements only by name and by 'any products other than food'.
Source: Guengerich 2026 (FEMA Expert Panel), Food and Chemical Toxicology Industry-funded Link to this fact
Hidden ingredients / Other evidence
Many chemicals on the founding 1965 FEMA flavor list are also common fragrance ingredients or labeled contact allergens, among them eugenol, isoeugenol, cinnamaldehyde, citral, linalool, geraniol, hydroxycitronellal, and d-limonene; the list covers eating them at the reported food levels, not breathing or skin contact.
Moderate evidence: FEMA numbers read from scanned OCR and checked on the page image only for eugenol and methyl eugenol; the paper does not discuss fragrance use.
Source: Hall & Oser 1965 (FEMA GRAS 3), Food Technology Industry-funded Link to this fact
Hidden ingredients / Other evidence
FEMA's 2009 list also affirmed sodium lauryl sulfate (FEMA 4437, 0.001 ppm), 2-phenoxyethanol (FEMA 4620, up to 8 ppm in baked goods) and camphor (FEMA 4513, up to 100 ppm in chewing gum) as flavoring substances, so a food 'flavor' can include chemicals better known from personal care products.
Moderate evidence: Industry's own use-level survey; food only; the usual and maximum levels are averages.
Source: Smith et al. 2009 (FEMA Expert Panel), GRAS Flavoring Substances 24, Food Technology 63(6), Food Technology Industry-funded Link to this fact
Hidden ingredients / Measured in people or real products
In 2010 lab tests of 17 popular fragrances (13 bought in the US, 4 in Canada) found 38 chemicals missing from the label of at least one product, about 14 per product (range 7 to 24); the highest was American Eagle Seventy Seven with 24.
Moderate evidence.
Caveats
NGO-commissioned report, not peer reviewed; one purchase per product, no replicates, no detection limits; most non-target compounds were tentative library matches; omission under 'parfum' was legal.
Source: Sarantis et al. 2010 (Not So Sexy, Canadian edition) Link to this fact
Hidden ingredients / Measured in people or real products
The lab was Analytical Sciences in Petaluma, California, using modified US EPA methods 8260 and 8270 (GC/MS) plus a separate GC/MS method for six synthetic musks.
Moderate evidence: Method description is brief; no QC data or confirmation status for each compound is given.
Source: Sarantis et al. 2010 (Not So Sexy, Canadian edition) Link to this fact
Hidden ingredients / Measured in people or real products
All 17 tested perfumes, colognes, and body sprays contained at least one synthetic musk; Galaxolide was in 15 and was not on any of those 15 labels.
Moderate evidence: Qualitative detection; musk concentrations are not reported; one sample per product.
Source: Sarantis et al. 2010 (Not So Sexy, Canadian edition) Link to this fact
Hidden ingredients / Measured in people or real products
Diethyl phthalate was found in 12 of the 17 fragrances, up to 32,000 ppm (about 3.2% by weight) in Calvin Klein Eternity for Women, and in none of the AXE, Bath & Body Works, Clinique, Dolce & Gabbana or Armani products tested.
Moderate evidence: Levels printed for only four products; 2010 formulations; small convenience sample.
Source: Sarantis et al. 2010 (Not So Sexy, Canadian edition) Link to this fact
Hidden ingredients / Review of other studies
Of the 38 secret chemicals, 10 had no toxicity study in PubMed and at least 6 more had three or fewer, including hedione, found in 16 of 17 products.
Moderate evidence: EWG's own PubMed search in 2010; absence from PubMed is not absence of industry data; counts of 'unassessed' ingredients are inconsistent within the report.
Source: Sarantis et al. 2010 (Not So Sexy, Canadian edition) Link to this fact
Hidden ingredients / Government agency
EU rules treat flavorings as ingredients (Art. 2(2)(f): "including flavorings"), and the exemptions from the ingredient list in Art. 20 apply "without prejudice to Article 21", so an Annex II allergen present in a flavoring must still be declared even though the flavoring itself may be listed as just "flavoring(s)".
Moderate evidence: This is a reading of several articles together; the regulation never states "allergens in flavorings" in so many words. Enforcement and compliance were not examined.
Source: EU Reg 1169/2011 (OJ L 304) Link to this fact
Hidden ingredients / Other evidence
The same panel cleared 560 previously GRAS flavors for higher use levels in chewing gum (year-2000 industry survey of 768 flavors covering over 95% of US gum), arguing that ppm in gum cannot be compared with ppm in other foods because a gum stick is chewed 20 minutes and only about 10% of a flavor is released.
Moderate evidence.
Caveats
Industry survey and the panel's own intake reasoning; the chew-out rate is from an industry program on 44 of the 560 flavors, and the full data are 'expected to be reported elsewhere'.
Source: FEMA GRAS 25 (2011), Food Technology Industry-funded Link to this fact
Hidden ingredients / Other evidence
Danish clothing companies mostly cannot say which dyes are in their clothes: of 18 firms answering, 13 had no written dye information, and only one named any dye.
Moderate evidence: Only 34 of 79 firms replied (43%), and non-response may reflect the same lack of knowledge; Danish firms only.
Source: Danish EPA 2014, Survey of Chemical Substances in Consumer Products No. 129 Link to this fact
Hidden ingredients / Other evidence
FDA laboratory testing in 2010 and 2016 found that Hyland's Baby Teething Tablets contained inconsistent amounts of belladonna, sometimes far exceeding the label claim (as reported).
Moderate evidence: Secondhand report; the primary FDA letter is on a sibling page.
Source: AVMA 2017 (JAVMA News) Link to this fact
Hidden ingredients / Other evidence
GRAS 28 gives usual and maximum use levels in ppm by food category for each new flavor, for example erythritol 5,000/12,500 ppm in non-alcoholic beverages, and updates use levels for 18 previously GRAS flavors (cinnamic acid at 700/1,000 ppm in milk products is a new level).
Moderate evidence: Industry-reported averages that the panel used for its judgments, not FDA limits; no ingredient labeling follows from them.
Source: FEMA Expert Panel 2018 (GRAS 28), Food Technology Industry-funded Link to this fact
Hidden ingredients / Other evidence
FDA noted some King Bio homeopathic drugs are made from Nux vomica, which contains strychnine, and said making them with an unvalidated process could expose patients to unnecessary risk.
Moderate evidence: FDA gives no strychnine amounts or dilutions; the concern is process control, not a measured dose.
Source: FDA warning letter to King Bio, 20 Mar 2019, FDA warning letter (20 March 2019) Link to this fact
Hidden ingredients / Other evidence
The two-page FASTER Act text never mentions flavors, spices or 'natural flavor'; whether sesame must be declared inside a flavor comes from the older allergen-labeling provisions and FDA's reading of them, not from this Act.
Moderate evidence: Absence from a 2-page statute; the flavor duty must be confirmed in FALCPA and FDA's Q&A pages.
Source: U.S. Congress, Public Law 117-11 (FASTER Act), 135 Stat. 262 (2021), Public Law 117-11, 135 Stat. 262 Link to this fact
Hidden ingredients / Measured in people or real products
The chamber test detected 530 distinct VOCs from 30 cleaning products and air fresheners, and only 17% of them appear in the cleaning industry association's ingredient database (American Cleaning Institute); 23 of 30 products were not fragrance-free and 7 of those listed just 'fragrance'.
Moderate evidence.
Caveats
EWG-authored (advocacy nonprofit that ranks cleaners and licenses EWG VERIFIED to cleaner makers); Skyline Foundation grant; unblinded; Detection of a chemical in chamber air does not show it was an intentional ingredient; some are impurities, by-products or sampling artifacts; 71% of VOCs were found in a single product.
Hidden ingredients / Other evidence
EWG Verified cleaning products must disclose every intentionally added fragrance ingredient publicly down to 0.01 percent, which no US federal rule requires of household cleaners.
Moderate evidence.
Caveats
Applies only to products that opt in and are licensed; the document states requirements, not compliance rates; amounts of ingredients are not disclosed. The comparison with US law relies on our other pages, not on this document.
Source: EWG Verified criteria for cleaning products (10.24), Environmental Working Group, 2024 Link to this fact
Hidden ingredients / Other evidence
FDA says some homeopathic products have been found to contain measurable amounts of their active ingredients, some of which are made from known poisons such as nux vomica (strychnine source), belladonna, mercury, and lead compounds.
Moderate evidence: Undated agency summary with no product names, amounts or counts on this page; product-level evidence is in the Hyland's recall letter and warning letters.
Source: FDA consumer page: Homeopathic Products, FDA web page Link to this fact
Hidden ingredients / Other evidence
The same panel kept tarragon (estragon) oil, which contains estragole, as GRAS after an industry-sponsored intake model pushed its margin of exposure above 10,000 (the first calculation was below it); basil, mace, nutmeg, and parsley oils also stayed GRAS.
Moderate evidence: Panel's own judgment and its own benchmark; the modeled intakes are not shown in the document.
Source: FEMA Expert Panel 2024 (GRAS 31), Food Technology Industry-funded Link to this fact
Hidden ingredients / Other evidence
US allergen labeling is triggered by protein: flavors, colors, and incidental additives derived from a major allergen but containing no protein from it are not subject to allergen labeling, so what a 'natural flavor' is made from is not itemized.
Moderate evidence: Exemption depends on the manufacturer demonstrating protein-free status; FDA notes lactose and ghee often carry residual milk protein.
Source: FDA food allergen Q&A Ed. 5 (2025), FDA guidance for industry (Docket FDA-2022-D-0099) Link to this fact
Hidden ingredients / Review of other studies
A 'natural' or 'organic' hair-dye label does not guarantee the product is free of PPD or toluene-2,5-diamine sulfate; the full ingredient list has to be read.
Moderate evidence: Secondary citation in a narrative review; no product counts given in this paper.
Information, not medical advice. See also: myths we won’t tell you.