"SPF: The Number Nobody Checks"

19 September 2026

A sunscreen labelled SPF 50+ tested at 4. Its brand then had it retested at several labs and got back eight different answers for the same product [3, 4]. So we asked a simple question: who actually checks the number on your bottle? In the United States, not the regulator. Independent testing often finds products below their label, and the amount you apply moves your protection more than the gap between SPF 30 and 50.

What we checked

What the label said

SPF measures less than most of us assume. It is a sunburn number: how much longer protected skin takes to burn than bare skin, and burning is mostly ultraviolet B. At the top of the scale the steps are small. SPF 30 holds back about 97% of UVB; SPF 50 about 98%.

The number is not read off a machine. It comes from a test on people. The US rule requires at least ten volunteers, sunscreen spread on their backs at 2 mg per square centimetre, a lamp at rising doses, and a technician who judges the next day where the skin turned pink [1]. A person applies it and a person reads it.

The label also leaves out UVA, the part of sunlight tied to ageing, which contributes to skin cancer too. In the US, "broad spectrum" is pass or fail: the product must hold its protection out to 370 nanometres, and that is the whole bar. In Europe and Australia, a product carrying the UVA symbol must also give UVA protection worth at least a third of its SPF. In one published comparison of 20 US sunscreens, 95% cleared the US rule and 55% met the European one [10]. A Korean PA++++ rating or a European UVA circle tells you something a bare American "50" does not.

What the evidence shows

Who tests. The FDA does not test sunscreen. It requires manufacturers to test their own products and keep the results, which they only send in if asked. Testing is required at launch or after a reformulation, not for every batch [1]. Before a test, the brand tells the lab what number it is hoping for, and if the result is low, it can go to a different lab.

How precise the number is. Dr Curtis Cole, who has represented the US for nearly 20 years on the committee that writes these standards, says there is no single "true" SPF; the test estimates an average. Even among the best labs, results for the same product have varied by as much as 64% either way [6].

What independent testers found. In Australia, the consumer group CHOICE sent 20 popular SPF 50 and 50+ sunscreens to an accredited lab; 16 came back under their label [3]. The brand with the worst result, Ultra Violette, had its product retested at several labs and got 4, 10, 21, 26, 33, 60, 61 and 64. It withdrew the product and refunded customers [4]. In the US, Consumer Reports tested 130 sunscreens this year and found most came in below their labelled SPF. Its top scorer was a $9 drugstore lotion for babies, so price is not buying accuracy [2].

Who shows their work. In April, BeautyMatter asked 36 top-selling US sunscreen brands for their test reports. Of those, 30 declined, ignored the request or sent the wrong document. Two sent reports that stood up to expert review: Beauty of Joseon and Badger [5].

None of this proves anyone lied. Results move between labs, brands dispute independent numbers, and a sunscreen that lands under its label still protects you, perhaps less than the front says.

The bigger gap is ours. The label is measured at 2 mg per square centimetre. Studies of real use put people at a fifth to a half of that [8]. Researchers at King's College London found that SPF 50 applied the typical way gave at best about 40% of its expected protection [9]. The drop may be gentler than the old textbook maths suggested, but it is real, and it is the part you control.

The verdict

The number on the bottle is an estimate from a test the manufacturer commissions, with real variation between labs: not fiction, and not a guarantee. We would buy Beauty of Joseon Relief Sun for daily face use, as one of only two brands whose report held up, with a PA++++ UVA rating on top. Badger is our pick for mineral: 25% zinc oxide, five other ingredients, and the other brand that showed its work. Eucerin's spray was the top-scoring spray in this year's Consumer Reports testing, with a caveat: spray until skin glistens, then rub in, because you cannot see how much you have used [2]. We would wait on Supergoop! Unseen. The texture is lovely, and liking a sunscreen is a real reason to wear it, but a proper face dose every morning costs nearly $1 a day, which encourages rationing, the habit that costs the most protection.

Buy broad spectrum, at least SPF 30, ideally with a UVA rating, and enough that you never ration it.

Worth knowing

• The tested amount is about a quarter of a teaspoon for the face and about a shot glass for the whole body. Reapply after two hours outside.
• Australia's regulator is consulting on replacing the numeric SPF with "low", "medium", "high" and "very high" [7].
• After CHOICE's test, 21 products were later recalled in Australia [3].
• This post is not medical advice. If you have concerns about your skin or sun sensitivity, talk to a clinician.
• No sponsorships. Nothing here was gifted or paid for.

Watch the video

The video sets out each test result with its source on screen: Check This Before Buying A Sunscreen - Most Failed Independent Lab Tests.

Sources

1. US FDA, Labeling and Effectiveness Testing: Sunscreen Drug Products (21 CFR 201.327)
2. Consumer Reports, "Best Sunscreens of 2026", 2026
3. CHOICE Australia, sunscreen test, 2025 (updated December 2025)
4. Ultra Violette, statement, August 2025
5. Janna Mandell, "Is Everyone Lying About Their SPF Claims?", BeautyMatter, April 2026
6. Dr Curtis Cole, "The Variability of In Vivo Sunscreen SPF Values"
7. TGA (Australia), consultation paper, March 2026
8. Petersen & Wulf, Photodermatology, Photoimmunology & Photomedicine, 2014
9. Young et al., King's College London, 2018
10. Wang, Xu & Stanfield, Journal of the American Academy of Dermatology

Every claim above is sourced in the text. Prices are what we paid, on the date shown. We buy what we test; nothing is gifted, and nothing is sponsored.