TL;DR
Seven ingredients with 2024-2026 evidence strong enough to retire, plus three the internet still panics about that I am keeping. The cut-list is shorter than most articles suggest. Slow skincare means fewer products, not more outrage.
I get tired of ‘toxic ingredients’ lists that recycle the same panic from a decade ago. Most of them are wrong. Some of them genuinely have a point. Sorting which is which takes about an hour with PubMed and the EU SCCS opinion archive open in two browser tabs.
Here is what I retired this year. Seven ingredients with real recent evidence. The reasoning is on each entry.
1. BHA as a preservative (not BHA the exfoliant)
Butylated hydroxyanisole, the synthetic antioxidant used to preserve fats in cosmetics. The EU SCCS reclassified it as CMR Category 1B in 2022. Added to the prohibited list in Annex II under Regulation 2022/2195. The US still permits it. I retire it on regulatory action; the carcinogenicity evidence in rodent models was the trigger. This is not salicylic acid. Different molecule entirely.
2. Methylisothiazolinone in leave-on products
A preservative with sensitization rates in patch-test cohorts that climbed to 10 to 13 percent at the height of its use. EU banned it in leave-on products in 2017. A 2023 review in Contact Dermatitis confirmed the allergy rate has stabilized but remains high enough to be the second most common contact allergen in cosmetics after fragrance. If a moisturizer contains MI in 2026, the formulator made a choice I cannot defend.
3. OTC hydroquinone
The FDA removed OTC hydroquinone from the GRASE list in 2020 under the CARES Act. Prescription use under dermatologist supervision is still legitimate for melasma. OTC formulas should not be sold. They still appear, especially on imported products. Long-term unsupervised use carries a real risk of exogenous ochronosis, which is permanent paradoxical hyperpigmentation. Skip the OTC versions.
4. PFAS in long-wear makeup
The Whitehead et al. 2021 paper in Environmental Science & Technology Letters found PFAS in 52 percent of long-wear cosmetics tested. EU phasing them out by 2026-2027. The persistence and bioaccumulation evidence is overwhelming. The direct skin-exposure risk is debated; the environmental case alone justifies the retirement.
5. Lilial
EU CMR 1B reclassification in 2022 under Regulation 2021/1902. Reproductive toxicity concerns from rodent studies. Being quietly reformulated out of global brands. Still legal in the US. I scan for it and skip products that contain it.
6. Retinyl palmitate in any daytime product
The molecule is fine at night. In daytime SPF or BB cream, photodegradation in UV produces compounds that may increase free radical load. NTP TR 568 (2012) found increased UV-induced tumors in mouse models. Human data is mixed but the conservative position is to keep retinyl esters off the morning routine entirely. If a daytime SPF lists retinyl palmitate in 2026, that is a formulation oversight.
7. Triclosan
Banned in OTC antibacterial soaps in 2016 by the FDA over endocrine disruption and antibiotic resistance concerns. Still in some toothpastes. Vanishingly rare in cosmetics now, but worth a check. The evidence on endocrine effects is solid.
The contrarian take: what I am keeping that the internet says to retire
Three things stay on the list. Parabens, because 60 years of safety data at cosmetic concentrations is hard to argue with, and the 2004 paper that started the panic has never been replicated. Mineral oil, because the cosmetic grade is highly refined and the carcinogenicity panic confuses it with industrial-grade contaminants. Silicones, because the chemistry on ‘skin suffocation’ was never correct in the first place. See our silicone deep dive if you want the full case.
Mixed feelings are allowed. You can prefer to avoid these. Preference is different from evidence-based avoidance, and the difference matters.
The real numbers on contact dermatitis rates
The North American Contact Dermatitis Group reported in Dermatitis 2021 that the leading allergens in skincare-related contact dermatitis were, in order: fragrance mix I (11.3 percent positive), methylisothiazolinone (7.3 percent), formaldehyde and releasers (6.4 percent), and balsam of Peru (6.0 percent). Parabens came in at 0.6 percent. PEGs at under 1 percent. Sulfates were not on the top-twenty list at all.
I find this gap between public worry and patch-test reality genuinely instructive. The things people fear and the things that actually cause allergic dermatitis are different things.
How to update your routine without drama
Use up what you have if you tolerate it. Check the INCI on your next purchase. Read the EU SCCS opinion archive once a year if you are the kind of person who enjoys regulatory documents. (I am.) Subscribe to one trustworthy dermatology newsletter. Ignore Instagram lists that did not cite anything.
For deeper label work, see our EU vs US regulation breakdown, the claim language audit, and the longer 2026 retirement list if you want nine entries instead of seven.
FAQ
Do I need to throw products out? Generally no. Use them up if you tolerate them. Replace at next purchase.
Is the EU automatically right? No. The EU uses a precautionary framework, which means some retirements there are based on lower evidence thresholds. The US is slower but not always wrong.
What about formaldehyde releasers? Real concern for the small percentage of formaldehyde-allergic people. Not a population-level retirement. If you have a known sensitivity, scan for DMDM hydantoin, quaternium-15, and similar.
Are ‘natural’ ingredients automatically safer? No. Essential oils are the second most common cause of allergic contact dermatitis in skincare after synthetic fragrance.
How often should I update this list? Once a year. The EU SCCS publishes opinions on a rolling basis. December is when I run the check.
Tag hub: More on skin science and evidence
Sources
EU SCCS Opinion SCCS/1636/21 on BHA. NACDG Patch Test Results, Dermatitis 2021. Whitehead HD et al. Environ. Sci. Technol. Lett. 2021. NTP TR 568 on Retinyl Palmitate. FDA OTC Monograph updates, 2016-2020.