Skip to content
Vanity or Vice

Every Skin in the Room

The Same Bottle Came Back as SPF 62 and SPF 5.5

Seven laboratories tested the same eight sunscreens under the same ISO standard. For one SPF 50+ product the answers ran from 62.4 down to 5.5. The endpoint every one of them was measuring is how pink a panel of skin goes.

Last updated

Evidence card: the claim "SPF 50 means fifty times the protection, and the number on the bottle is a measurement." set against the cohort that was actually measured.

Seven contract research organizations were sent eight sunscreens sold as SPF 50 or 50+, plus three reference products, and asked to run the standard test. The same protocol, ISO 24444. The same products.

Some came back with tight agreement. One did not. Its SPF was measured at 62.4 by one laboratory and 5.5 by another.

Not 62 and 45. Sixty-two and five.

One sunscreen, two laboratories, SPF 5.5 and SPF 62.4A sun protection factor scale running from 0 to 70. The same marketed sunscreen, tested under the same ISO 24444 protocol, was measured at SPF 5.5 by one contract laboratory and SPF 62.4 by another. The product is sold as SPF 50 or 50+, marked on the scale for comparison.One bottle. Seven laboratories. One protocol.Sun protection factor returned for the same marketed product010203040506070sold as SPF 50 / 50+5.5one laboratory62.4another laboratoryAcross the proficiency scheme running since 2009, dispersion between laboratories runs between 10 and 50 percent.Trullàs C, Granger C, Lim HW, Krutmann J, Masson P, Photodermatol Photoimmunol Photomed 2020;36(5):351-6.
The same bottle, the same ISO protocol, and an answer that moves by more than a factor of ten depending on who ran the test.

How the number is made

SPF is a ratio of two doses of ultraviolet light. The dose that just barely reddens a patch of protected skin, divided by the dose that just barely reddens a patch of unprotected skin next to it.

Both halves of that ratio are a person’s skin turning pink, and somebody deciding, by eye, that it has.

The bottom half of the fraction — the minimal erythema dose of unprotected skin, written MEDu — is the one to watch. The scientist who assembled the international dataset on it puts it as bluntly as anyone could want: MEDu is perhaps the most critical aspect of the SPF value, and it disproportionately influences the result.

Disproportionately, because it is the denominator. Get it slightly wrong and the whole factor moves.

So how does a lab know a subject’s MEDu before it starts? Since the 2019 revision of the standard, it estimates it from their pigmentation. A dataset of 9,931 individuals from twelve laboratories on five continents was regressed against individual typology angle — the colorimetric measure of skin lightness — and that curve now sits inside ISO 24444 as the way to set a provisional MEDu for a subject nobody has tested before.

I want to be fair about this, because it is the field doing something right. Using a measured color value rather than asking somebody to guess their Fitzpatrick number is exactly the correction this section keeps asking for, and it was done for a good reason: to stop labs from wandering off with wildly different MEDu assumptions.

It also tells you something. You do not build a pigmentation-to-dose regression into a standard unless pigmentation moves the answer.

Who was measured Transfers in part

SPF 50 means fifty times the protection, and the number on the bottle is a measurement.

Source
Trullàs C, Granger C, Lim HW, Krutmann J, Masson P, Photodermatol Photoimmunol Photomed 2020;36(5):351-6; Cole C, Colson B, Uhlig S, Int J Cosmet Sci 2025;47 Suppl 1:25-36; Zago DI, Ben Bari S, Tirard A, et al., Int J Cosmet Sci 2024;46(6):1097-104; Granger C, Trullàs C, Sokeechand NB, et al., Photodermatol Photoimmunol Photomed 2024;40(1):e12942
Participants
8 marketed products across 7 laboratories; 32 products across 12 laboratories; 26 laboratories in an annual proficiency scheme running since 2009; 334 subjects across 50 studies at a single test center
Measured
Panels of human volunteers whose skin is irradiated until it reddens. The 2019 standard sets each subject's starting dose from a regression on 9,931 individuals' individual typology angle.
Not measured
Any published breakdown of SPF test-panel composition by skin tone. The endpoint is visible erythema, and no alternative endpoint is codified.

In 334 subjects at one center, individual typology angle and MEDu both influenced the SPF value obtained. Age and sex did not.

A ratio of two rednesses, read by eye, reproducible to within a factor of ten.

The spread is not one bad laboratory

It would be comforting if that 62-versus-5.5 result were a single rogue site. It is not, and the industry has been measuring its own scatter for years.

A French proficiency-testing scheme has been sending identical samples to laboratories around the world since April 2009 — twenty-six of them, one round a year, scored against the international statistical standards for this kind of comparison. Its published summary describes a relatively constant dispersion of data, running between 10 and 50 percent.

A larger 2025 exercise built specifically to benchmark new non-invasive methods sent 32 products in eight groups to twelve laboratories. Its finding on the old method: variability between laboratories is proportional to the SPF, so the highest-factor products are the least reproducible, and repeatability within a single laboratory is much better than agreement between them.

The authors’ own conclusion is the sentence to keep. The interlaboratory variability of SPF results is considerable and is likely greater than expected by the public and regulatory bodies.

By the public. That is us.

What else moves it

A single testing center in Bucharest pooled 50 studies run on the two ISO reference sunscreens between 2021 and 2022 — 334 subjects, 276 observations on the SPF 50+ reference alone.

Age did not affect the result. Sex did not affect the result. Season did. MEDu did. And individual typology angle did.

In principle the ratio should cancel pigmentation out: if darker skin needs more ultraviolet to redden, that dose appears on the top and bottom of the fraction. In practice, in that dataset, it did not fully cancel. Skin color was one of the variables shifting the number that ends up on the label.

There is also a formulation-side artifact worth knowing if you buy mineral sunscreens. High-concentration zinc oxide products come out lower on the newer in vitro tests than they do in vivo, and the likely reason is arithmetic rather than chemistry: the standards dose the test plate by weight, zinc formulations are much denser than emulsions of organic filters, and the same weight therefore spreads a thinner film. Preliminary work applying the same volume instead of the same weight brings the two methods back into line. Which is to say a zinc sunscreen may test badly for a reason that has nothing to do with how it performs on a face.

The one comparison against actual sunlight

Buried in the seven-laboratory study is the part I find most useful, because somebody took the products outdoors.

The reason some sunscreens scattered so badly was that laboratories were using different irradiation dose regimens, and products divided into two behaviors: those whose measured protection stayed constant across regimens, and those whose protection changed with the dose used to test them.

Five of the products were then taken to Mauritius and tested on people under high natural sunlight.

The two categories did not differ in preventing sunburn.

So the laboratory disagreement was, at least in that experiment, an artifact of the test rather than a real difference in protection — and the direction of the error was that low-dose laboratory testing underestimated some products. That is a strange kind of reassurance. The number is unreliable, and its unreliability is not simply consumers being oversold.

Somebody has proposed giving up on the number

In 2026 Brian Diffey, who has spent a career on this, published a proposal to abolish SPF labeling and replace it with a binary category — protected to a threshold, or not.

His stated reason is the uncertainty in both the in vivo and in vitro methods. When a senior figure in a field proposes deleting its central metric, that is a fair summary of how much confidence the metric deserves.

How much meaning the digits carry

None of this is an argument against wearing sunscreen every day. The mechanism is not in doubt; the outcome trials in richly pigmented skin are a separate and emptier story, but the filters absorb what they are documented to absorb.

What changes is how much meaning to load onto the digits.

Treat SPF as a rank, not a measurement. A 50 is more protective than a 15. The gap between a 30 and a 50 sits comfortably inside the scatter that twelve laboratories produce on the same product, so choosing between them on the number alone is choosing on noise.

Spend the attention you were giving the number on the amount instead, because application quantity moves real-world protection far more than the label does, and most people apply a fraction of the test dose. And on whether it covers what you actually need covered — for pigment-prone skin the relevant question is visible light, which the SPF number does not measure at all.

The SPF on a bottle is the average of a redness judgment made on a panel of volunteers, in a laboratory whose answer for that same bottle might be a third of another laboratory’s. It is not a lie. It is just a much softer number than it looks, and it was never built to describe your skin in particular.

Evidence check: Strong

Evidence level: interlaboratory variability has been measured repeatedly by the industry itself, under a proficiency scheme running since 2009 and in a twelve-laboratory benchmarking exercise. The skin-tone effect on the result is one factor among several in a single-center pooled analysis.

Sources reviewed September 14, 2026: Trullàs C, Granger C, Lim HW, Krutmann J, Masson P, “Linear and exponential sunscreen behaviors as an explanation for observed discrepancies in sun protection factor testing,” Photodermatol Photoimmunol Photomed 2020;36(5):351-6, PMID 31376288. Cole C, Colson B, Uhlig S, “The variability of in vivo sunscreen sun protection factor values,” Int J Cosmet Sci 2025;47 Suppl 1:25-36, PMID 40637051. Zago DI, Ben Bari S, Tirard A, Miksa S, Renoux P, Questel E, “Overview of proficiency testing results for the in vivo determination of sun protection factor,” Int J Cosmet Sci 2024;46(6):1097-104, PMID 39051146. Cole C, “Global data of unprotected skin minimal erythema dose relationship to Individual Typology Angle,” Photodermatol Photoimmunol Photomed 2020;36(6):452-9, PMID 32654351. Granger C, Trullàs C, Sokeechand NB, Jourdan E, Krutmann J, Francois-Newton V, “Evaluating the factors influencing sun protection factors (SPF): Pooling data from multiple studies involving two reference ISO 24444:2019 sunscreen products (P2 and P8),” Photodermatol Photoimmunol Photomed 2024;40(1):e12942, PMID 38288771. Osterwalder U, Hubaud JC, Perroux-David E, Moraine T, van den Bosch J, “Sun-protection factor of zinc-oxide sunscreens: SPF in vitro too low compared to SPF in vivo – a brief review,” Photochem Photobiol Sci 2024;23(10):1999-2009, PMID 39432202. Diffey B, “A proposal for simplifying the labeling of sunscreens,” Int J Cosmet Sci 2026, PMID 42037070.

This article is educational and is not a substitute for care from someone who can look at your skin. Anything new, changing, painful or spreading should be assessed rather than researched.