Tea Tree Oil Is in Seven of Our Products. The EU Just Called It a Moderate Sensitiser.

Seven of the fifty-six products we sell have tea tree in them. We counted twice, because the number of products moved on us: we had fifty-four a week ago, and two hair-colour items landed since. The tea tree count held at seven.

Tea tree has one of the calmest reputations in the ingredient list. It reads as the gentle option, the botanical you reach for when benzoyl peroxide has stripped your chin raw. It is the kind of ingredient that gets printed on the front of the bottle rather than buried at line forty-one. One of ours has it in the product name.

On 30 October 2025 the EU's Scientific Committee on Consumer Safety adopted an opinion that describes it, in its own words, as a moderate skin sensitiser.

What the committee actually decided

The opinion is not a ban and not a clearance. It is a table of four numbers.

Twenty-fold between the top and the bottom of that table, and the logic behind the spread is exposure rather than potency. A shampoo is on your scalp for ninety seconds and then goes down the drain. A face cream sits there until you wash it off tomorrow. Same molecule, same skin, different dose over time.

Two conditions came attached. The oil is cleared for adult products only, and the opinion states it should not be used in aerosolised or sprayable delivery formats. Both of those are easy to skim past and neither is decorative.

It is worth putting the 2025 opinion next to the one before it. In December 2008 the committee's predecessor looked at the same oil and concluded that a margin of safety had not been calculated and the safety of tea tree oil could not be assessed. That was the official European position for seventeen years. The 2025 document is the first time anyone in Brussels put a number on a face cream.

The problem is not tea tree. It is old tea tree

Hausen and colleagues worked this out in 1999, and the finding has held up better than most things published that year. They aged tea tree oil and tested what came out.

Peroxide content went from under 50 ppm to over 500 ppm in four days. The p-cymene fraction climbed from about 2% to 11.5%. Fresh oil behaved as a very weak sensitiser in their model; the oxidised material was roughly three times stronger. Eleven patients already sensitised to tea tree were patch tested against the individual breakdown products, and every one of them reacted to alpha-terpinene, terpinolene and ascaridole.

The sensitising agent, in other words, is not really in the bottle when the bottle is filled. It arrives later, out of light and air. Ascaridole is not something anyone adds to a formula.

We should be careful about how hard we push that. The sensitisation work in Hausen's paper ran on guinea pigs; the human portion was a patch test on people who were already allergic. So it establishes what the culprits are, not how often a shopper becomes allergic in the first place.

For that second question the best source is de Groot and Schmidt's 2016 review in Contact Dermatitis, which put the positive patch test rate in routine testing somewhere between 0.1% and 3.5% and counted roughly a hundred published cases since 1991. Their observation that matters most for anyone reading an ingredient list: the clinical reactions mostly came from people applying pure oil, not from formulated cosmetics. The standard patch test material is 5% oxidised tea tree oil, which is a concentration no face cream would ever contain.

Two acne trials, thirty-six years apart

The reputation rests on two trials.

Bassett, Pannowitz and Barnetson ran 124 patients in the Medical Journal of Australia in 1990, comparing 5% tea tree gel against 5% benzoyl peroxide lotion. Both worked. Tea tree was slower to show an effect and produced fewer side effects. That is the whole result, and it gets misquoted in two directions: as evidence tea tree beats benzoyl peroxide, which it did not, and as a double-blind trial, which it was not. It was single-blind, and for an endpoint as subjective as lesion counting that matters.

Enshaieh and colleagues published the placebo-controlled version in 2007 in the Indian Journal of Dermatology, Venereology and Leprology: 60 patients, thirty on 5% tea tree gel and thirty on placebo, followed for 45 days. Tea tree came out 3.55 times better than placebo on total lesion count and 5.75 times better on acne severity index.

Those multiples look enormous. They are ratios against a placebo arm, so a small placebo response inflates them without anything happening to the treatment arm, and the paper does not publish the underlying counts in its abstract. We went looking for the p-values that get quoted alongside these numbers all over the internet and could not find them in anything we could actually open. So we are not printing them either.

Sixty people. Forty-five days. One centre. A reasonable bet, not a settled question.

The number nobody prints

Here is what makes the SCCS table hard to use in a shop.

It is a table of percentages, and Korean cosmetic labelling does not require percentages. Neither does European labelling, for that matter, outside a short list of specific substances. The ingredient list is ordered by descending weight down to 1%, and below that the order is free. So when the committee says 0.1% in a face cream, there is no way for a shopper — or for us — to confirm whether a given product sits under that line.

We went looking for concentration data on our own seven. We do not have it. Our suppliers do not publish it and we have not asked for it in a form we could verify. That is a gap in our own method and we would rather write it down than work around it.

What our screening found

The practical takeaway

Buy small bottles and finish them. Almost everything in the sensitisation literature is about oxidation, and oxidation is a function of time and headspace. A 200 ml bottle of anything with essential oils in it that lives on a sunny shelf for two years is a different product at the end than at the start.

Treat neat oil as a separate category from formulated products. The hundred-odd published allergy cases cluster around people dabbing undiluted oil on a spot, which is roughly fifty times the concentration the committee thinks belongs in a face cream.

If you already react to fragrance, be more careful here than the botanical framing suggests. The sensitisers are terpene breakdown products, out of the same chemical family that fills most of the EU's fragrance allergen list.

And do not read a mid flag from us as a verdict on the ingredient. On this one it was a verdict on the shelf it sat on. We have written before about the ten ingredients we flag most often, about the alcohol-free claims in our own catalogue that do not survive a reading of the ingredient list, and about a product we sell whose name describes an ingredient it barely contains. The scoring method, including the parts we are in the middle of rebuilding, is on our awards page.