The Ten Ingredients We Flag Most, and How Worried You Should Actually Be

Nine of the products on this site contain salicylic acid. Twelve contain something from the fragrance family. We know because we typed all of it into a database, one INCI name at a time, and the database does not care who owns the product.

Last month we published the method. This is the output: the ten ingredients that trip our flag most often across the catalogue, ranked by how many products they appear in, with the actual evidence behind each flag laid out so you can decide whether our worry matches yours.

Some of these flags deserve less alarm than the word "flagged" implies.

One of them turns out not to be about your skin at all.

The shape of the data before the list

Fifty of our fifty-four products have a complete ingredient list we could parse. The other four don't publish one we could get at, which is its own small scandal.

We can't fix that from this end.

The distribution matters more than the ranking does. When 90.5% of everything we read comes back low-risk, a "flagged" ingredient is not a warning label. It's a note in the margin saying this one has a literature attached to it. Eight entries out of 1,696 got our worst grade, and every one of those eight sits in a product we sell, which is a piece of arithmetic we'll be publishing in full on its own soon.

Fragrance, and the number nobody quotes

Twelve products. It's the biggest single block of flags we carry, and the reason is structural rather than chemical: "fragrance" on a label can legally stand in for dozens of individual substances, and the formulator is not obliged to tell you which. You're being asked to accept a redaction.

The rate people repeat for fragrance allergy comes from patch-test clinics, where everyone present already has a rash. That's the wrong denominator. Diepgen and colleagues did it properly in 2015: a random general-population sample across Germany, Italy, the Netherlands, Portugal and Sweden, 12,377 people interviewed, 3,119 of them patch tested. Fragrance mix I came back positive in 2.6% (95% CI 2.1–3.2), fragrance mix II in 1.9% (1.5–2.4). Their conservative estimate of clinically relevant fragrance contact allergy was 1.9%, or 0.8% under the strict definition. Women run about double the rate men do.

So: roughly one person in fifty, and about one in a hundred if you're strict. That's a real number and not a trivial one. It is also not the "everyone should avoid fragrance" number that the wellness internet implies, and we'd rather you had the honest version.

Linalool and limonene: it's the oxygen, not the molecule

These two are why the fragrance conversation is more specific than it sounds.

Fresh linalool and fresh limonene are close to inert. What sensitises people is autoxidation: leave the bottle open, let air do its work, and hydroperoxides form that are genuinely potent haptens.

The size of that gap is startling once you see it side by side. Audrain's UK study patch tested 4,731 consecutive patients across 13 dermatology departments. Oxidised limonene: 237 positives, 5.0%. Oxidised linalool: 281 positives, 5.9%. Reacting to the stabilised, non-oxidised form alone: 11 patients. 0.2%. Testing the oxidised forms alone caught 411 of the 422 positives, or 97.0%. The molecule on the label is not the molecule doing the damage.

Here's the part we didn't expect to find. EU Annex III regulates both, at the same declaration thresholds of 0.001% in leave-on and 0.01% in rinse-off products. But the limonene entry carries one extra line that the linalool entry does not: "The peroxide value for each substance shall be less than 20 mmoles/L." Limonene's oxidation state is capped by law. Linalool's is not, despite Audrain finding more positives to oxidised linalool than to oxidised limonene. We read the consolidated regulation twice looking for the linalool equivalent. It isn't there.

One more thing worth knowing if you buy in Europe or from European-facing brands. Regulation 2023/1545 expanded the individually-declarable list by 56 substances. The compliance date for new formulations was 31 July 2026, which was nine days ago. Existing stock has until 31 July 2028, so labels will disagree with each other on shelves for the next two years.

Salicylic acid, tea tree and menthol: three different kinds of flag

Salicylic acid sits in nine of our products and it is the flag we're least worried about, which sounds like special pleading until you look at the exposure arithmetic. The Cosmetic Ingredient Review's assessment notes that roughly 10% of applied salicylates stay in the skin, and put daily use of a representative cosmetic product at about 20% of the systemic exposure you'd get from a single 81mg baby aspirin. The EU caps it at 2.0% in most products, 3.0% in rinse-off hair products and 0.5% in body lotion plus several eye and lip products, with a hard exclusion for children under three except in shampoo. Our flag exists because those caps exist and because the total across a stacked routine is a number nobody calculates. Not because 2% on a patch is dangerous.

Tea tree oil is the opposite case: a flag that earns its weight. Rutherford's Melbourne review found 41 positive patch tests to oxidised tea tree oil out of 2,320 people tested, a 1.8% prevalence, and judged the reaction clinically relevant in 17 of the 41. Twenty of those people had been applying the oil neat, at 100%. Freshly opened tea tree oil, the same paper notes, produces no or weak reactions. So this is the linalool problem again, wearing a different hat.

Menthol we flag for a reason most people get wrong. It isn't the cooling. The issue sits in the Cosmetic Ingredient Review's peppermint oil assessment, which caps peppermint oil at 3% in rinse-off and 0.2% in leave-on formulations, limits its pulegone content to 1%, and then adds the line that actually matters: menthol enhances the penetration of other agents, and the panel warned formulators this "can affect the use of other ingredients whose safety assessment was based on their lack of absorption." A penetration enhancer changes the risk profile of everything else in the bottle. That's why it's flagged. We also went looking for a standalone CIR monograph on menthol itself and there isn't one, only the peppermint oil document, so don't let anyone cite you a menthol monograph that doesn't exist.

The flag that shouldn't be a skin flag

The practical takeaway

Read a flag as a pointer to a literature, not as a verdict. Eight of the ten entries above are mid-grade, which in our system means the ingredient has published concerns attached at some concentration, in some population, under some storage condition. Whether any of that reaches you depends on which of those conditions you're actually in.

If you have known fragrance sensitivity, the useful move is not avoiding "fragrance" on labels. It's avoiding old bottles. Linalool, limonene and tea tree oil are all near-harmless fresh and meaningfully allergenic oxidised, so the citrus-scented cleanser that's been open on a sunny windowsill for two years is a different product from the one you bought.

And treat any brand's ingredient grading, ours included, as a set of choices rather than a measurement. We built a risk column that mixes environmental persistence with skin irritation and didn't notice until we wrote this piece. You can see exactly how the grading works, why most label claims carry less weight than the grading does, and how the salicylic acid question plays out in the one category where we compared ourselves to the US bestsellers. Every product that clears the screen is listed on our awards page, flags and all.