At the centre of the SPF number is a patch of human skin.
In the standard in-vivo SPF test, sunscreen is spread onto small marked areas at two milligrams per square centimetre. A solar simulator exposes those areas to ultraviolet light. Hours later, trained staff look for the minimum dose that produces just-perceptible redness, then compare protected skin with unprotected skin. The Australian Therapeutic Goods Administration describes the ISO 24444 method as a human test usually run on ten to twenty healthy adults, with the redness reading made roughly sixteen to twenty-four hours after exposure.1 The ISO sample describes the same basic architecture: a standardized artificial UV source, defined product application, and SPF calculated from the ratio between protected and unprotected minimal erythemal doses.2
That is not a fake number. It is the way standardization has tried to turn sunburn into a label claim.
It is also not a physical constant. It is an estimate built from people, skin response, laboratory practice, product spreading, and a statistical rule. The bottle turns that estimate into one clean integer. The shelf does not show the confidence interval.
The wrong conclusion is the easy one: the whole category cannot be trusted. The evidence does not say that. A randomized trial in Nambour, Australia followed 1,621 residents and found that daily SPF15+ sunscreen reduced the number of squamous-cell carcinoma tumours compared with discretionary use.12 NHS and FDA public guidance still tell people to use broad-spectrum sunscreen alongside shade, clothing, hats, and sensible timing.1113 The question is what kind of accountability sits behind the exact number printed on the bottle.
The Skin Test
SPF is a sunburn number. More precisely, it is a UVB-weighted erythema number measured under laboratory conditions. A product is applied much more evenly and thickly than most people apply it in real life. It is then judged against redness, not against skin cancer directly, UVA ageing, waterproofness, user behaviour, or whether a beach bag has cooked the bottle for a week.
That is why an SPF50 label can sound more precise than it is. TGA's consultation explains the non-linear scale plainly: SPF30 filters about 97% of UVB under test conditions, while SPF50 filters about 98%. A product claiming SPF50+ must demonstrate an SPF above 60. A product that measured 55 would still offer high UVB protection, but it would not meet an SPF50+ claim under that scheme.1
The public conversation often collapses those two facts. Claim accuracy and protection are not the same thing. A claimed SPF50 product that tests around SPF30 has not become useless. It has become a different promise.
The 2025 paper that matters most here did not test a single headline product. It tested the test. In "The variability of in vivo sunscreen sun protection factor values," Cole and co-authors analysed thirty-two sunscreen products across twelve laboratories. They found that between-laboratory variability increased with SPF and was "considerable" enough that the authors expected it would surprise the public and regulators.3 The paper identifies familiar friction points: subjective erythema grading, the definition of the minimal redness dose, and the way small judgement differences expand when converted into an SPF value.3
That is the shelf problem. The consumer sees "50" as if it is a weighed kilogram. The system that produced it knows it is closer to a laboratory estimate.
The Buyer Of The Judgment
The accountability chain begins before the bottle reaches a shopper, but it does not usually begin with a public regulator repeating the test.
In Great Britain, sunscreen is handled through the cosmetic-products framework. A Responsible Person must be in place. The product must have a Product Information File, a safety assessment, and a notification submitted to the Office for Product Safety and Standards before it is made available to consumers.4 The Responsible Person must also be able to show that claims are correct; GB guidance says labels and advertising must not imply characteristics or functions the product does not have, and authorities can require corrective action, withdrawal, or recall.5 The retained EU claim framework also says efficacy claims should be truthful, honest, supported, and useful for an informed decision.6
That is real accountability. It is not the same as a public pre-sale audit of every SPF number.
Australia's system makes the private-evidence issue unusually visible because TGA is now consulting on it. The consultation says listed sunscreen sponsors must hold evidence that the SPF claim was established by testing, but the SPF data are not provided to TGA at listing and do not appear in the public ARTG entry. The sponsor chooses the laboratory. TGA sees the underlying data when it asks for them, not as a routine condition of the number reaching the shelf.1
The consultation then says the part a shopper cannot see: current oversight gives limited visibility over which laboratory was used, how rigorously testing was performed, whether the final market-ready batch was tested, and how often a product should be retested after scale-up or shelf-life changes.1
That is the accountability gap. The label makes a public number. The evidence behind that number is often private.
After-Shelf Alarms
Independent testing shows why that gap matters, and also why it must not be exaggerated.
In September 2025, Which? reported testing fifteen sun creams sold in the UK. Two did not meet its tests: one for SPF and one for UVA performance. Thirteen passed. Which? also reported that low-priced products from supermarkets and high-street retailers passed, which is an important check against a lazy premium-versus-budget story. The same article says Which? retests failures and buys products as a consumer would before sending them to independent labs.8
That is not a category collapse. It is a visible failure mode inside an otherwise functioning category.
Australia supplied the sharper case. CHOICE reported that sixteen of twenty SPF50+ sunscreens it tested did not meet their labelled SPF, with one product first reported around SPF4 and retested around SPF5 at another laboratory.9 TGA's consumer update described the same public concern more cautiously: CHOICE had found sixteen of twenty products did not meet SPF50+ claims, one was reported at SPF4, seven were above SPF20, and eight were above SPF30. TGA told consumers not to stop using sunscreen and said it was investigating.7
The recall record matters because it cuts both ways. It shows the problem was not imaginary; TGA later recorded recalls of multiple sunscreens where preliminary evidence indicated SPF may be lower than claimed.10 It also shows the system can respond after evidence surfaces. A post-market recall is accountability. It is just not the same as confidence at the moment of purchase.
This investigation continues below.
Want the next one in your inbox?
No named company needs to be turned into the villain for this to be useful. The exposed pattern is structural: a small, variable, sponsor-paid test is converted into a clean public integer, while the raw assurance normally sits behind the counter.
The Better Check
The technical path out is no longer hypothetical.
TGA's consultation points to two newer ISO methods published in 2024: ISO 23675, an in-vitro SPF method, and ISO 23698, a hybrid diffuse-reflectance method. The consultation does not treat them as a magic replacement. It notes limits, including water-resistance claims and validation across the full product range. But it also explains why they matter: they may reduce reliance on deliberate human UV exposure and make batch or periodic testing more feasible than a full in-vivo panel every time.1
That is the practical hinge. If SPF can be checked more often, at lower ethical and financial cost, then the number can become less like a one-time certificate and more like a controlled specification.
The reform that would change the buyer's confidence is not a warning label saying "do not trust sunscreen." It is a narrower disclosure architecture: which method was used, which laboratory category performed it, whether the final market formulation was tested, when it was last checked, and whether a regulator can see the data before a problem becomes public. A recurring batch-test record would do more for the SPF number than another marketing word on the front of the bottle.
The Useful Number
For the person standing in the aisle, the best reading is practical rather than cynical.
The SPF number is useful because it distinguishes broad levels of UVB sunburn protection under controlled conditions. It is not useful as a promise that a state laboratory has rechecked that exact bottle before sale. The NHS advice is still the boring one: choose at least SPF30 with strong UVA protection, use enough, apply before going out, reapply at least every two hours, and do not rely on sunscreen alone.11 FDA says the same in American language: sunscreen helps protect skin when used with other sun-protection measures, broad spectrum matters, no sunscreen is waterproof, and reapplication matters.13
The interval on the bottle matters most when a buyer is close to the edge. A person who applies half the tested quantity, misses ears and neck, swims, sweats, and waits four hours between applications has already moved far outside the lab. In that world, the difference between an accountable SPF50 and an unaccountable SPF50 is not academic. It is the difference between a number behaving like a specification and a number behaving like a hope.
Evidence that would narrow this analysis is easy to name. A public OPSS or TGA-style pre-market SPF data requirement would narrow it. Routine publication of method, laboratory identity, formulation status, and last-test date would narrow it. Repeated batch or shelf-life testing with the newer ISO methods would narrow it. A large, transparent inter-laboratory programme showing stable high-SPF reproducibility across finished products would narrow it.
Until then, the SPF number should be read as a useful regulated estimate, not a public guarantee. Wear the sunscreen. Ask more of the number.