Titanium Dioxide
Titanium dioxide is rated Clear, and for once the circulating number is a real measurement: Fulton's 1989 rabbit-ear assay scored it 0, the same 0 he gave the other inert mineral pigments beside it, and his 1984 paper prints 0–1 for it, which that paper's own key files as not significant. The two readings agree, but they come from the same investigator's programme rather than from independent laboratories. What they do not settle is a mineral sunscreen that broke someone out — an ingredient's score says little about the finished formula it sits in.
Fulton 1989 scored 0 and Fulton 1984 0-1 of 5 — two readings, one programme.
What it is
An inorganic mineral, titanium dioxide (TiO2), used in cosmetics as a white pigment and opacifier and as a broad-spectrum mineral UV filter. On a label it is Titanium Dioxide, or CI 77891 when it is doing the job of a colour (its pigment name is Pigment White 6). It is insoluble in water, and its CAS number is 13463-67-7. One word of care, because it is the crux of the page: "titanium dioxide" is a single INCI name covering physically different materials. It is sold in two crystal forms (anatase and rutile, the rutile form typical in sunscreens), in a large pigment grade that reads as opaque white and a micronized or nano grade fine enough to be near-transparent, and — for cosmetic use — it is usually surface-coated to make it disperse and stay photostable. Those coatings are themselves ingredients: alumina, silica, dimethicone or simethicone, and fatty-acid treatments such as stearic acid, isostearic acid or polyhydroxystearic acid. A single commercial UV-filter grade can carry three or four of them at once. So the powder Fulton put on a rabbit's ear and the coated micronized grade in a modern sunscreen share a name and not much else.
Why we rate it Clear
We rate it Clear at Low confidence. The evidence starts, as almost every comedogenic number does, with Fulton's 1989 rabbit-ear assay — the study nearly every ingredient list quietly quotes. Fulton diluted ingredients to roughly 10% in propylene glycol, applied them to the inner ear of New Zealand albino rabbits, three rabbits per assay, five days a week for two weeks, and measured the follicular keratosis under a micrometer. His table has an exact titanium dioxide row, and it reads 0. By his own published key, a grade of 0 to 1 "is not considered significant." This is worth stating plainly because it is unusual on this site: a great many circulating comedogenic numbers turn out to trace to nothing, or to a different material, or to a user comment. Titanium dioxide's 0 is a reading Fulton actually took, of the material he actually named. There is a second Fulton reading, and it carries its own key. The 1984 paper he wrote with Pay and Fulton prints "Titanium dioxide* — 0–1" in Table I, under "Miscellaneous pigments", on a 0–5 scale whose thresholds are not the 1989 ones: there, "a minimal grade of 1 to 2 is not considered significant", and 3 or above is considered positive. The asterisk means something different in that paper too — it marks a sample diluted in mineral oil or made up in an alcohol gel at 10%, not a result depending on the source of the raw material. So 1984 reads not significant and 1989 reads 0, and the two agree. They are two readings from the same Fulton programme, five years apart, and not independent replication by a second laboratory. Neither reading sits alone in its own table. In the same section of the 1989 table, the other inert mineral pigments scored the same: iron oxides 0, carmine 0, ultramarine violet 0. Several of the synthetic organic dyes listed beside them scored 2 or 3. The 1984 pigment block reads the same way, with ultramarine violet, iron oxides and carmine all at 0–1. That is a description of how the mineral colourants read in those two tables, and we leave it there: it is not a further experiment on titanium dioxide, and it does not raise our confidence in the 0. Confidence stays Low for that reason. Two legacy rabbit-ear readings from one investigator, on samples he does not describe, are legacy-assay evidence rather than a consensus. There is no mechanism we can cite that has been tested in a follicle, and "inert mineral" is a description, not an experiment.
What the evidence doesn’t tell you
The 0 and the 0–1 are rabbit-ear readings, from the same assay we treat with caution everywhere else on this site. The rabbit ear is not a face, Fulton called his own survey "not at all definitive," and the 2006 re-evaluation by Draelos and DiNardo found that rabbit-ear scores do not reliably transfer to human skin. A 2025 review in JAAD Reviews, surveying the field from 1972 onward, found that no standardised comedogenicity test exists at all and that the label "non-comedogenic" is unregulated. Those cautions apply to a 0 exactly as they apply to a 4. We identified no human comedogenicity study of titanium dioxide, and no comedogenicity measurement of it from a laboratory other than Fulton's in the sources we searched: Morris and Kwan's 1983 rabbit-ear study does not test it, and neither does Nguyen 2007. Content farms on this topic circulate a specific clinical trial — a 0.5% titanium dioxide preparation on thirty young women with facial acne — with no author, journal or year attached; we could not trace it to a primary document and so we do not use it. That leaves the honest state of the evidence: two concordant readings from the same Fulton programme, and no independent replication we could find. Fulton did not record which titanium dioxide he tested in either paper — no crystal form, no particle grade, no coating, no supplier. Given that the name covers pigment grade, micronized and nano grades, and a range of surface coatings that are themselves separate ingredients, his unspecified samples cannot stand in for all of them. Nor does either paper record the physical form the material was in: 1989 says nothing about it, and the 1984 asterisk marks the sample as diluted in mineral oil or gelled in alcohol at 10%. Neither reading is a test of the coated micronized grade a sunscreen actually uses. And the score is about the ingredient, not the product. Draelos and DiNardo's point cuts both ways: a finished formula can test comedogenic despite low-scoring ingredients, and it can test clean despite high-scoring ones. When a mineral sunscreen breaks someone out, a low number for the mineral itself does not clear the formula it is carried in — the emollients, waxes and film-formers that make a heavy mineral paste spreadable are also on that label, and no assay we hold looked at them together. We can say titanium dioxide the mineral scored 0 in 1989 and 0–1 in 1984. We cannot tell you the product on your shelf will not break you out, and nothing in the evidence we hold identifies what caused a given breakout. Finally, a distinction that is routinely blurred. The safety questions raised about titanium dioxide — the EU's scientific review of nano grades, and the occupational limits on inhaling the powder — are about inhalation and about specific nanomaterial coatings, not about pores. Safety and comedogenicity are different questions, and an inhalation concern is not evidence about a follicle in either direction.
Where you’ll see it
On labels as Titanium Dioxide, or CI 77891 when it is listed as a colour. It is one of the most common ingredients in the pigment and sun-care aisles: mineral and hybrid sunscreens, foundations and concealers, BB and CC creams, pressed and loose powders, primers, and lip products, plus countless creams where it simply supplies opacity and a white base. In the European Union it is permitted as a UV filter up to a maximum of 25%. Its position on an ingredient list tells you little on its own — it can be a headline UV filter at high percentage or a trace opacifier — and the score on this page is for the mineral, not for whatever else shares the label with it.
Sources
Each source says what it actually is. Several widely-cited “sources” for comedogenic ratings are republishing the same 1989 assay, and counting them as independent agreement is how a thin evidence base gets made to look thick.
- Fulton JE. "Comedogenicity and irritancy of commonly used ingredients in skin care products." J Soc Cosmet Chem 40:321–333 (1989) The primary source for the 1989 reading, and one of the two Fulton papers in the corpus we searched that tested titanium dioxide. A rabbit-ear model: ingredient diluted to roughly 10% in propylene glycol, applied to the inner ear of New Zealand albino rabbits, three rabbits per assay, five days a week for two weeks, follicular keratosis measured with a micrometer. Read here off the transcribed table rather than a list quoting it. Titanium dioxide has its own row, scored 0; the neighbouring inert mineral pigments (iron oxides, carmine, ultramarine violet) are also 0, while several synthetic D&C dyes in the same section score 2–3. Fulton's own grade key: 0–1 "is not considered significant." He records no crystal form, particle grade, coating or supplier for the sample, and calls the whole survey "not at all definitive but simply designed to stimulate research."
- Fulton JE Jr, Pay SR, Fulton JE III. "Comedogenicity of current therapeutic products, cosmetics, and ingredients in the rabbit ear." J Am Acad Dermatol 10(1):96–105 (1984) The earlier Fulton rabbit-ear survey, read off a 400dpi render of the journal scan. Method: two New Zealand albino rabbits per test, the material applied to the inner surface of one ear daily for two weeks with the other ear as control, follicular hyperkeratosis graded 0–5 and borderline cases checked by microscopic biopsy. Table I prints "Titanium dioxide* — 0–1" under "Miscellaneous pigments", beside ultramarine violet, iron oxides and carmine at the same 0–1. Two things belong to this paper and not to the 1989 one. Its key: "a minimal grade of 1 to 2 is not considered significant", with 3 or above considered positive. And its asterisk, which marks a material diluted in mineral oil or incorporated into an alcohol gel at 10% — not the 1989 asterisk's source-of-raw-material caveat. A second reading by the same investigator, concordant with 1989, and not an independent replication.
- Draelos ZD, DiNardo JC. "A re-evaluation of the comedogenicity concept." J Am Acad Dermatol 54:507–512 (2006) A Current Issues and Opinion piece that tested finished products on human upper backs and found products testing non-comedogenic despite containing ingredients the rabbit assay had scored comedogenic. Cited for the caveat it actually establishes — that an ingredient's score does not transfer cleanly to the formula it ends up in — which is why a 0 for the mineral does not clear the sunscreen it sits in. It is not a study of titanium dioxide and not a validation of the rabbit assay against human skin.
- "Comedogenicity in cosmeceuticals: a review of clinical relevance, regulatory gaps, and future directions." JAAD Reviews (2025) A recent peer-reviewed review of the whole field, covering literature from 1972 to 2025. Finds that no standardised comedogenicity test exists and that "non-comedogenic" is an unregulated label. A synthesis of the legacy literature, not a new measurement, and independent of Fulton only in authorship, since the corpus it surveys includes his. Cited for the field-wide caveat.
- SpecialChem Cosmetics INCI directory. "Titanium Dioxide (CI 77891)" (updated 16 Jan 2023) A supplier-facing trade reference, not a study and not a comedogenicity source — cited only for material identity, which it documents concretely. It gives the chemistry (TiO2, CAS 13463-67-7, CI 77891 / Pigment White 6, insoluble, white), the two crystal forms (anatase and rutile), the fact that cosmetic titanium dioxide is typically micronized and surface-coated, the surface modifiers used (stearic acid, isostearic acid, polyhydroxystearic acid, dimethicone/methicone copolymer, and — in a listed commercial grade — alumina, silica and simethicone), the functions (opacifier, UV filter, colorant), and the EU's 25% maximum for use as a UV filter. Its safety commentary concerns inhalation of the powder, not the skin; its loose consumer-blog phrasing ("non-risky") is not evidence and is not relied on here.
Others in the same family
Minerals & pigments behave similarly enough that the evidence for one is often wrongly read across to the others. These are its structural relatives, not a guess. What the evidence says about minerals & pigments as a family ›
Last reviewed 2026-08-12 · How we decide