How we decide
This is a checker for one question only: how likely a product’s ingredients are to clog pores (comedogenicity). We don’t claim scientific authority — comedogenicity has thin, contested evidence. What we offer instead is a transparent, opinionated read that shows its sources and its confidence on every call.
What each rating means
Every rated ingredient gets one of three editorial ratings: Clogging (commonly linked to clogged pores), Caution (mixed or concentration-dependent), or Clear (no comedogenic concern on record). Ingredients we recognize but haven’t rated show as “recognized — no concern on record,” which is neutral to the verdict.
Confidence comes from the evidence
Confidence is derived from the strongest evidence behind a rating, never set by feel: a human or clinical study gives High confidence, dermatological consensus gives Moderate, a legacy lab assay alone gives Low, and a rating that rests only on someone’s uncited list — no assay, no study, no stated method — also gives Low, and we label it as exactly that. The old 0–5 comedogenicity scores you may have seen elsewhere come from rabbit-ear assays; we record them for transparency but never let them drive a verdict on their own.
A rating can also carry no legacy score at all. Sometimes that means the ingredient was never tested by anyone — which is not the same thing as being tested and found clean, and we refuse to record it as though it were. Shea butter is the clearest example: it has never appeared in any comedogenicity study we can find, so its score is blank rather than zero.
Where those 0–5 numbers actually come from
This is the part almost nobody says out loud, so we will.
Nearly every comedogenic rating on the internet — including the ones we record — descends from a single 1989 paper by James Fulton in the Journal of the Society of Cosmetic Chemists. Each ingredient was diluted to roughly 10% in propylene glycol, applied to the inner ear of New Zealand albino rabbits, three rabbits per assay, and the follicles were scored for keratosis. Rabbit-ear testing was already in use before that paper, but Fulton’s table is the one the lists copy: it is where the numbers you have read come from.
And they get copied badly. The most-quoted comedogenic score in skincare is “coconut oil is a 4” — and there is no coconut oil in Fulton’s table. He tested coconut butter. A number measured on one material has been attached to another for thirty-odd years, and every list repeating it, including ours until we actually read the paper, simply passed the error along. We have corrected it, and said so on the coconut oil page.
It is not a human face. It is not a finished formula. It is not a realistic concentration. And the objection isn’t ours — it’s published. A 2006 re-evaluation in the Journal of the American Academy of Dermatology found that rabbit-ear results do not reliably predict what happens on human skin, and that testing an isolated ingredient says little about the product it ends up in. A 2025 review in JAAD Reviews covering the literature from 1972 onward reached the same conclusion, and added that there is still no standardised test for comedogenicity at all — which is why any brand may print “non-comedogenic” on a bottle without having measured anything.
So when a site tells you an ingredient is “a 4” and stops there, it is quoting a rabbit from 1989 with a confidence the study cannot support. We record the same number, label it a legacy assay, cap its confidence at Low, and refuse to let it produce an Avoid by itself. That constraint is the entire reason this tool exists.
What we mean by “nobody has tested it”
Several ingredients on this site carry no legacy score, and we say so on their pages. It would be easy, and it would be wrong, to write that as nobody has ever tested it. That is a claim about every paper ever published, and we have not read every paper ever published.
We know exactly how wrong it can go, because it has already gone wrong here. Chasing sodium lauryl sulfate’s famous 5, we found a ninth numbered paper — Fulton 1976, in Cutis — that our own reference file had never heard of. It had been sitting in the footnotes of a 1983 industry report for forty years. Every time we have said “there are only N documents,” N has turned out to be wrong.
So instead of a claim about the world, here is a claim you can check: this is the corpus we searched. When a page says an ingredient has never been assayed, it means it does not appear in these documents. That is a smaller statement, and a true one.
- Kanaar P. Follicular-keratogenic properties of fatty acids in the external ear canal of the rabbit. Dermatologica 142(1):14-22 (1971) kanaar-1971
- Kligman AM & Mills OH. 'Acne cosmetica.' Arch Dermatol 106:843-850 (1972) DIFFERENT SCALE. A 3 here is the MAXIMUM, not a middling score. kligman-mills-1972
- Kligman AM & Kwong T. Br J Dermatol 100:699-702 (1979) DIFFERENT SCALE. kligman-kwong-1979
- Mills OH & Kligman AM. Arch Dermatol 118:903-905 (1982) DIFFERENT SCALE. One of the very few papers with HUMAN data. mills-kligman-1982
- Morris WE & Kwan SC. J Soc Cosmet Chem 34:215-225 (1983) 32 rows: 7 lanolins, 9 vegetable oils, 11 esters, 5 surfactants. Its Table III (oils) is printed ROTATED and OCR returns NOTHING for it — anyone who scraped this paper by OCR lost the entire oils table. It openly CONTRADICTS Fulton on several materials and says so. morris-kwan-1983
- Fulton JE, Pay SR, Fulton JE 3rd. J Am Acad Dermatol 10(1):96-105 (1984) Tested finished acne PRODUCTS, cosmetics AND ingredients. What the read changed: (1) SODIUM LAURYL SULFATE = 5 — the flat 5 the charts print is Fulton's own 1984 reading, not 'invented'; see retraction sls-5. (2) ACETYLATED LANOLIN = 4, with NO separate 'acetylated lanolin alcohol' row (that split is a 1989 refinement) — Fulton's two papers conflict on the plain ester (4 in 1984, 0 in 1989); see retraction acetylated-lanolin-4. (3) CETYL ALCOHOL = 3 (vs 2 in 1989). Corroborations of 1989: myristyl myristate 5, isopropyl palmitate 4, glycerin 0, propylene glycol 0, isopropyl myristate 5, isopropyl isostearate 5, decyl oleate 3, oleyl alcohol 4. The abstract NAMES octyl stearate, octyl palmitate and isocetyl stearate as offenders but gives them NO numeric row (prose classification, not a measured score). fulton-1984
- Lanzet M. Cosmet Toilet 101:63-72 (1986) DIFFERENT SCALE. lanzet-1986
- Fulton JE. J Soc Cosmet Chem 40:321-333 (1989) The 220-row table nearly every online score descends from. See data/fulton-1989-table1.json. fulton-1989
- Nguyen SH, Dang TP, Maibach HI. Cutan Ocul Toxicol 26:287-292 (2007) DIFFERENT SCALE AGAIN (0-4). No per-material score is readable from the abstract, so no Nguyen number may be carried under D56 — only the positive/not-positive classification above. nguyen-maibach-2007
- Francis A & Shojan A. 'Comedogenicity of Oils.' Int J Contemp Med Res 6(8):H5-H7 (2019) NOT part of the classic corpus and close to worthless as evidence. IJCMR appears by name on Beall's List of predatory journals; 22-day submission-to-publication; 4 of 24 rabbits died with no necropsy; the authors ACCEPTED THE NULL HYPOTHESIS on n=4-6 ears ('Since the p-value in each case is greater than 0.05, we accept the null hypotheses'); liquid paraffin, their own comparator, produced comedones in 3 of 5 ears; there was NO untreated ear anywhere in the study; and the paper acknowledges funding from Marico (maker of Parachute, one of the test articles) while its footer declares 'Conflict of Interest: None'. CITE IT SO A READER WHO FINDS IT KNOWS WHAT IT IS. Do not use it as evidence, and do not pretend it does not exist. It tested virgin coconut oil, mill coconut oil, Parachute, mustard oil and LIQUID PARAFFIN (i.e. mineral oil). francis-shojan-2019
Plus, where they exist, the industry safety assessments for the ingredient in question (the Cosmetic Ingredient Review’s final reports — and we mean the final ones; we have been caught citing three separate comment-stage drafts as though they were settled), and the published human studies named on each page. If it is not in that set, we have not seen it. That is not the same as it not existing, and we are not going to write it as though it were.
We keep a list of the numbers that are wrong
Coconut oil is not the only one. We have now found 12 widely-repeated comedogenic claims that no document supports, and each one has a paper behind it that says otherwise. Acetylated lanolin’s famous 4 belongs to a different molecule, and Fulton says so himself, in the same paper. Sodium lauryl sulfate’s 5 belongs to nothing at all: two rabbit assays tested it, got opposite answers, and the internet resolved the disagreement by inventing a number higher than either of them.
Who is behind this
One person makes every rating call on this site. Her name is Krista. She is not a dermatologist, not a cosmetic chemist, and not a laboratory, and nothing here should be read as though she were. There is no institution behind this, no peer review, and no lab bench. It is one person reading old papers carefully.
That is a thin claim to authority, and we would rather state it plainly than dress it up, so here is the honest version of what the expertise actually consists of: we read the primary document, and we show you what it says. Not a list that quotes a paper. The paper. When we write that Fulton’s table has no coconut oil row in it, that is because the table was rendered at 300dpi and read. When we cannot get a document, we say we could not get it, and we do not rate as though we had. That is the entire method, and it is the only thing on offer here that most of the sites with better credentials are not doing.
Everything else on this page is the machinery that keeps one person honest. Every page is handed to an external adversarial fact-check, with the primary documents attached and instructions to assume we are wrong, before it is allowed a URL. Every claim we have killed is written into a ledger the build checks every page against, so a dead claim cannot come back. Neither of those is a credential. They are a substitute for one, and they are auditable, which a credential is not.
If something here is wrong, hello@porecelain.com is the fastest way to say so, and we would rather hear it from you than find it ourselves.
The verdict rule
The verdict is set by the single worst ingredient, then softened by its confidence. A Clogging ingredient with High or Moderate confidence makes the product Avoid. A Clogging ingredient backed only by a low-reliability legacy assay, or any Caution ingredient, makes it Caution. If everything recognized is Clear, we say No flags found — deliberately, because that is exactly what we determined, and nothing more. We don’t call it “safe”: an ingredient list can’t tell us how a product is formulated or at what concentration, so a clean list is not a promise about your skin. We also ignore how many flagged ingredients there are and where they sit in the list, because neither is reliable from an ingredient list alone.
Your allergies override everything
If you list a known allergy and we find it in an ingredient list, we mark the product Avoid — regardless of how it scores on pore-clogging. This is the one place where something other than comedogenic evidence sets the verdict, and we say so plainly on the result: we tell you the allergy drove it, and we still show what the product rates on comedogenicity alone, so you never lose that read. An allergy match carries no confidence level, because it isn’t an evidence claim — it’s your own filter, applied faithfully.
Allergy matching is a simple text search, so it is deliberately generous: typing “lavender” matches “Lavender Oil” and “Lavandula Angustifolia (Lavender) Oil” alike. It can over-match, and it can miss an allergen hiding under a name you didn’t think to type. It is an aid, not a medical safeguard — if you have a serious allergy, read the label yourself. Your allergies are stored only on your device and are never sent to us.
When sources disagree
Source lists often conflict. When they do, we flag it with a “sources disagree” badge that’s independent of confidence, and we explain the disagreement in the ingredient’s details. That conflict signal is a feature, not a hedge.
Recognized vs. not recognized
Every result shows how many ingredients we flagged, recognized, and couldn’t recognize, and how many ingredients we checked yours against. If we can’t recognize anything you pasted, we say so plainly rather than reporting No flags found — that’s an input problem, not a clean product. Tokens we can’t place are logged anonymously (the ingredient string only, never your full paste) so common ones can be added over time.
Sources & attribution
Ratings are our own arrangement of facts drawn from public ingredient lists — acne.org, CosDNA, Face Reality, Sofie Pavitt, Banish — with peer-reviewed studies as a tie-breaker where they exist. Each rated ingredient lists its contributing sources.
One honest caveat about that list, because it would be easy to make it look stronger than it is: those are not five sources agreeing with each other. They are, for the most part, one 1989 assay quoted five times. Several of them say so themselves. Counting them as independent corroboration would be exactly the sleight of hand this page exists to call out, so we don’t: a rating supported only by republishers of the legacy assay stays at Low confidence no matter how many of them agree. Confidence rises only when genuinely independent evidence — a human study, or dermatological consensus that isn’t simply restating the assay — actually exists.
Ingredient-name recognition uses the EU CosIng database. INCI names © European Union, reused under CC BY 4.0.
Not medical advice
This tool is informational only and not medical advice. Whether an ingredient actually causes breakouts depends on its concentration and the finished formulation, which an ingredient list can’t reveal. If a product matters for your skin or health, talk to a dermatologist.