Aluminium in Deodorant: What the Evidence Actually Shows

You'll do this roughly 18,000 times across an adult lifetime — once or twice a day, fifty years across an adult lifetime. The active ingredient is almost always aluminium chlorohydrateAn aluminium salt formed from aluminium chloride and aluminium hydroxide. The most common antiperspirant active ingredient — typically 15–25% by weight in the product., and you've applied it within centimetres of breast tissue, on skin you've just abraded with a razor. So is aluminium in deodorant something to worry about?

The honest answer has shifted twice in the last decade. The mainstream cancer agencies — ACSAmerican Cancer Society and NCINational Cancer Institute — say there is no evidence of a breast cancer link. Germany's BfRBundesinstitut für Risikobewertung — the Federal Institute for Risk Assessment reversed its own 2014 warning in 2020. A 2024 meta-analysis pooling seven case-control studies found no association. And yet France Avoid on damaged skin 2011 still applies, the SCCSEU Scientific Committee on Consumer Safety only signed off concentrations after capping them, and the in vitro mechanism work in Sappino 2012 is real. This article walks through what's settled, what isn't, and what to do — part of our evidence-based guide to endocrine disruptors.

Aluminium in deodorant is an aluminium-salt active ingredient — almost always Al₂(OH)₅Cl aluminium chlorohydrate or the zirconium-stabilised form aluminium zirconium tetrachlorohydrex gly — that blocks sweat by forming a temporary plug at the opening of the eccrine sweat duct. It is what makes a product an antiperspirant rather than a deodorant. USUnited States regulators classify antiperspirants as over-the-counter drugs, not cosmetics, because of this physiological action (21 CFR 350).

The mechanism was worked out in detail in 2017 by a team using microfluidics and small-angle X-ray scattering to watch what happens when an aluminium-chlorohydrate solution meets sweat proteins. The aluminium polycations and the proteins aggregate together. The aggregates grow, lodge inside the sweat duct, and physically obstruct flow Bretagne et al. 2017. The plug stays put until the stratum corneumThe outermost dead-cell layer of skin. It naturally sheds and renews on a roughly monthly cycle. sloughs the duct lining off — usually one to several days. The body never opens the plug; the skin grows out around it.

It matters that an antiperspirant is doing something physiological. A deodorant — a perfumed alcohol or a triclosan-style antimicrobial — masks odour or kills the bacteria that turn fresh sweat into the smell. It does not block the gland. If the label doesn't list an aluminium compound near the top of the active ingredients, the product is a deodorant. That distinction is doing more work in this article than any single statistic.

Antiperspirant gets the headlines because the application is daily, the area is intimate, and the chemistry is overt. But a typical daily dose of aluminium reaches the body from several routes at once, and the antiperspirant route is — by mass — usually the smallest of them.

A single 600 mg antacid tablet contains roughly the same elemental aluminium that the strongest antiperspirant on the EUEuropean Union market would deliver topically across years — but the antacid goes through the gut, where gastrointestinal absorptionHow much of an ingested substance crosses the intestinal lining into the bloodstream. For aluminium, this is roughly 0.1% under normal conditions — so even large oral doses translate to small absorbed doses. of aluminium runs around 0.1% under normal conditions. The deodorant question is less about quantity and more about route, repetition, and where the application happens — applied to the underarm, every morning, under shaved skin, beside the lateral edge of the breast.

The honest answer in 2026 is: the available human evidence does not support the claim, and the strongest summary so far — a 2024 meta-analysis pooling seven case-control studies — found no association (OR 0.96 95% CI 0.78–1.17 — Cancer Investigation 2024 meta-analysis of seven case-control studies of antiperspirant use and breast cancer risk, 95% CI 0.78–1.17, Allam et al. 2024). The American Cancer Society's October 2022 review concludes there is "very little scientific evidence" supporting a link. The National Cancer Institute's fact sheet is the same. The hypothesis began as a serious one and has narrowed under three decades of testing.

What kept the hypothesis alive long enough to test thoroughly was a real biological signal. The framing was set out by Philippa Darbre in a 2005 hypothesis paper that argued underarm cosmetics deserved investigation as a contributor to breast cancer because of dermal exposure, repeated application, and proximity to breast tissue Darbre 2005. Subsequent measurements found something. Concentrations of aluminium in human breast tissue measured by Exley and colleagues in 2007 ranged from 4 to 437 nmol/g dry weight, with the outer regions — axilla and lateral — significantly higher than the inner regions — middle and medial (p=0.033) Exley et al. 2007. Mannello's 2009 work on women with gross cystic breast disease found aluminium in apocrine (type I) breast cyst fluid at a median of 150 μg/L — about 25 times the median in serum (~6 μg/L) Mannello et al. 2009. And in an in vitro study using MCF-10AA non-cancerous human mammary epithelial cell line widely used in breast biology research. normal mammary epithelial cells, long-term low-dose aluminium chloride at 10–300 μM produced loss of contact inhibition, anchorage-independent growth, and DNA double-strand breaks Sappino et al. 2012. The Sappino authors were explicit: their findings "do not formally identify aluminium as a breast carcinogen" but "question the safety ascribed to its widespread use in underarm cosmetics."

Three things were observable. Aluminium reaches breast tissue. It accumulates more on the side closest to the underarm. And it can do plausibly carcinogenic things to mammary epithelial cells in a dish. The next question was whether any of that translates into measurable risk in actual women.

The first widely-cited human study was a 2003 retrospective survey of 437 women who had already been diagnosed with breast cancer. McGrath asked them about lifetime antiperspirant use and underarm shaving, and reported that women who used antiperspirants more frequently and started earlier were diagnosed at younger ages McGrath 2003. There was no control group. The exposure data was self-recalled by women already living with cancer. The methodology gap was wide enough that even McGrath's own conclusion called for proper case-control follow-up. The paper got cited a lot anyway.

The case-control follow-up McGrath asked for came the year before, in 2002. Mirick and colleagues at the Fred Hutchinson Cancer Research Center in Seattle compared 813 women newly diagnosed with breast cancer against 793 randomly selected age-matched controls. Antiperspirant use returned an odds ratio of 0.9 antiperspirant ever-use vs no-use; Mirick 2002, Seattle case-control n=1,606; OR 0.9, P=0.23 — no association (P=0.23). Deodorant use: OR 1.2 (P=0.19). Neither was statistically significant. The conclusion: "these findings do not support the hypothesis that antiperspirant use increases the risk for breast cancer" Mirick et al. 2002.

The one major positive case-control to follow was Linhart and colleagues' 2017 study from the Medical University of Innsbruck. They matched 209 Tyrolean breast cancer cases against 209 age-matched controls. The headline finding was real but specific: in the subgroup of women who used underarm cosmetic products several times daily and had started before age 30, the odds ratio was 3.88 — with a 95% confidence interval of 1.03 to 14.66 Linhart et al. 2017. The interval barely clears the line of significance, and the subgroup itself contained 27 patients. Subsequent peer-reviewed critiques flagged the small subgroup, the recall bias inherent to lifetime self-report, and the non-probability control sampling. The paper is the strongest single positive result in the literature, and the confidence interval is wide enough to drive a regulator's car through.

By 2024 the literature was mature enough to pool. Allam's meta-analysis combined the Mirick 2002, Linhart 2017, and five other case-control studies — with the negative findings outweighing Linhart's positive subgroup once the studies were combined on equal footing. The pooled odds ratio sat at 0.96 (95% CI 0.78–1.17). The authors noted what every honest reviewer notes about this literature: a prospective cohort study — one that enrols healthy women, asks about exposure now, and follows them for cancer over decades — has never been done, and would change everything.

The amount of aluminium that crosses intact human skin from an antiperspirant is small — and the best estimate has been falling, not rising, as the methodology has improved. The number that the deodorant-and-cancer literature was built on came from a 2001 study using two human subjects and a single underarm application of ²⁶Al-labelled aluminium chlorohydrate (84 mg topical dose, monitored by accelerator mass spectrometry over seven weeks). The fraction absorbed: 0.012% Flarend et al. 2001. That number has been quoted in regulatory and consumer literature for two decades.

A larger and more rigorous follow-up arrived in 2018. de Ligt and a TNO Netherlands team gave twelve healthy women between 23 and 39 a topical 25% ²⁶Al-aluminium-chlorohydrate formulation (matching the 6.25% elemental Al that the EU's safety committee considers safe), normalised against an intravenous dose of ²⁶Al-citrate to anchor each woman's individual urinary excretion baseline — the most rigorous methodology published to date. The mean fraction absorbed: 0.0094%, range 0.002–0.06% de Ligt et al. 2018. Funded entirely by Cosmetics Europe. Worth knowing.

Then de Ligt's 2022 follow-up dropped the absorption estimate again. Using the same ²⁶Al microtracer approach but with a topical dose 25 times higher than the 2018 study (which made the blood and urine signal cleanly measurable rather than at the limit of detection), the best estimate of the absorbed fraction came in at 0.00052% de Ligt 2022 ²⁶Al microtracer follow-up — about 23× lower than the 2001 figure that's been cited for two decades — about 23× lower than the 2001 number. de Ligt et al. 2022. Germany's BfR considers the 2022 paper methodologically superior to all prior work and treats it as the current best estimate.

The intact-skin number is one half of the absorption picture. The other half is what happens when the skin is not intact. In 2011 France's medicines agency — then AFSSAPS, now ANSMAgence nationale de sécurité du médicament et des produits de santé — France's medicines and health products safety agency — commissioned in vitro work on a 20% aluminium chlorohydrate formulation applied to Franz diffusion cellsA standard in vitro setup using excised skin mounted between two chambers — the formulation applied on top, a receptor fluid below. Used to estimate dermal absorption without testing on living subjects.: through intact skin, 0.5% absorbed; through stripped skin (the upper barrier layer mechanically removed), 18% absorbed. The agency issued a recommendation: cap cosmetic aluminium at 0.6% and avoid use on damaged or freshly-shaved skin. Pineau and colleagues confirmed the same direction the following year using human skin from five donors, with the stripped-skin uptake six times higher than intact-skin uptake from a stick formulation Pineau et al. 2012.

The numbers don't all line up neatly. The in vitro models on excised skin (AFSSAPS, Pineau) report higher absorption than the human radiotracer studies on living skin (de Ligt 2018, 2022). The most parsimonious read is that excised skin is more permeable than living skin, and that the freshly-shaved-skin caveat — which French regulators preserved through the SCCS reassessment and which BfR did not walk back — is the one regulatory-level guidance that all the data still supports. The dead-skin number is the one ANSM regulated on. The living-skin number is the one BfR walked the warning back on.

The aluminium-and-Alzheimer's hypothesis dates to a 1973 paper in Science that reported elevated aluminium in the brains of patients who had died with Alzheimer's-like neurofibrillary degeneration Crapper et al. 1973. It launched fifty years of measurement, animal work, and population studies. The strongest individual epidemiological signal came from France's PAQUID cohort: a 15-year follow-up reported that daily aluminium intake of ≥0.1 mg/day from drinking water was significantly associated with increased dementia risk, while geographic exposure to Al was not Rondeau et al. 2009. The water-source signal is robust; the broader hypothesis is not.

By 2014 the scientific community had largely abandoned the aluminium hypothesis as the primary causal explanation for Alzheimer's disease — though it persists in public discourse, partly because the actual causes of AD remain poorly understood Lidsky 2014. The dialysis encephalopathy literature — patients with severe renal failure exposed to high intravenous aluminium loads — is a separate, real, and well-characterised toxicity. It has very little to say about consumer-level dermal antiperspirant exposure. The Alzheimer's question is mostly closed for the antiperspirant route. The breast question is the one worth talking about.

Five jurisdictions, five different positions, all currently in force. The split is partly about science and partly about regulatory category — the United States treats antiperspirants as over-the-counter drugs because they alter a body function, while the EU and UK treat them as cosmetics with concentration caps.

Two of those rows do almost all the rhetorical work. EFSAEuropean Food Safety Authority's 2008 weekly tolerable intake of 1 mg Al/kg body weight is roughly half what JECFAJoint FAO/WHO Expert Committee on Food Additives, the WHO's expert body on the same chemical, set as its provisional weekly tolerable intake in 2011 (raised from the same 1 mg figure JECFA had previously used). The two leading food-safety authorities in the world disagree by a factor of two on what's safe. On the same chemical. With the same data.

The IARCInternational Agency for Research on Cancer classification deserves its own line because it gets misread constantly. IARC Monograph Volume 100F evaluates occupational exposures during aluminium production — the smelter environment, where workers breathe coal-tar pitch volatiles and polycyclic aromatic hydrocarbons during the anode-baking step. Those PAHs are the identified Group 1 carcinogens. Aluminium-the-element has not been evaluated by IARC as a standalone agent. Articles that say "aluminium is IARC Group 1" are quoting the metal's job title, not its actual classification.

The science has moved against the original cancer hypothesis. SCCSScientific Committee on Consumer Safety 2020 says cosmetic aluminium does not significantly add to body burden. BfRBundesinstitut für Risikobewertung 2020 reversed its 2014 warning. The 2024 meta-analysis returns OR 0.96. The 2022 human radiotracer data put intact-skin absorption at five-thousandths of a percent. The honest summary is that the strong claim — antiperspirants cause breast cancer — is not supported.

The weak claim is the one worth taking seriously. No prospective cohort has been run; the available evidence is case-control plus mechanism plus tissue measurement, which is exactly the configuration that historically produced both real signals (BPA, DEHP) and noise (saccharin, cell-phone EMF). The Linhart 2017 outlier exists. The Sappino in vitro signal is real. The Exley 2007 finding that breast tissue concentrations are higher on the underarm side has not been explained — accumulation, drainage, or coincidence are all live possibilities. The freshly-shaved-skin caveat sits on actual data and is regulator-supported. The repeated daily exposure for fifty years has never been studied prospectively in humans because nobody is ethically going to randomise that trial.

The practical math is what makes this an easy switch even with weak evidence. A no-aluminium deodorant — magnesium-based, baking-soda-based, a salt crystal, or just an alcohol-and-essential-oil stick — costs the same as the antiperspirant. It does a different job (covers smell rather than blocking sweat), so the trade is wetter underarms in exchange for stepping out of the uncertainty. There is no health upside to the antiperspirant. The downside, if there turns out to be one, is the one we can't yet measure. Watch

Switch because the alternatives are cheap, not because the alternatives are proven safer. That's the precautionary case in one sentence. It's the same case the brand applies to parabens where the headline studies don't replicate but the mechanism is biologically plausible — and the same case it does not apply to SLS, where the evidence is clear that the chemical is an irritant rather than an endocrine disruptor and switching is genuinely optional. The honest precautionary call depends on what the evidence actually says. For aluminium in antiperspirants, the evidence says the cancer claim is weak but the long-term-exposure question is open, and the alternative is free.

You're back at the bathroom mirror. The decision is one twist of the cap and a label glance away. If the active ingredient is an aluminium compound, you're looking at the antiperspirant the cancer literature mostly absolves and the absorption literature suggests is much less of an exposure than thought a decade ago. If you'd like to step out of the question entirely while it remains open, the swap costs the same and works on the smell. The only thing the evidence says with confidence is that you're allowed to make either choice without it being the one that matters most for your hormones. That one's still fragrance.