Reading the evidence

Skin purging runs on a number that was corrected in 1972

You are told to wait 4 to 6 weeks, because skin renews every 28 days. That 28 is not a measurement. It is a 1965 addition sum, and the field revised it to 40 to 56 days seven years later. The correction has been public and free to read ever since, and consumer skincare has never picked it up.

Nobody has measured how long purging lasts, because medicine does not have the word.

Written by Maya Faour How we score →Published 5 Aug 2026Reviewed 5 Aug 2026

The short answer

The discomfort is real. The timetable is not.

Is skin purging a real medical thing?: The experience is real, the word is not medicine's. A search of the whole biomedical literature returns two papers, neither about skin. Dermatology's word is retinization.
How long does it last?: Nobody has measured it. The 4 to 6 weeks you read everywhere was derived from a cell-turnover figure of 28 days, not from anyone watching faces.
Where did 28 days come from?: A 1965 paper added its own 13 days to 13 or 14 taken from another study. A 1972 review of the whole literature concluded the real figure was probably 40 to 56 days.
When is it actually worst?: The trials score irritation, not new spots. That peaks in the first week on the face and declines from there as tolerance develops.
Is purging different from breaking out?: No published rule separates them. A 2026 guideline advisory says flare-ups after starting topical retinoids "are not commonly documented in published literature".

If you were prescribed your retinoid, or your skin is painful, that is a conversation for your clinician. Here is when to see a dermatologist.

Medicine does not have a word for purging

We searched the entire biomedical literature through Europe PMC for the exact phrase "skin purging". It returns two papers.

Result 1

A systematic review of Lonicerae Japonicae Flos

A plant used in traditional Chinese medicine. Front Pharmacol, 2022.

Result 2

A half-yearly report on forensic medicine and toxicology

Published in 1857. Neither paper is about skin.

Search for purging alongside acne or retinoids instead and you get two Chinese patent formulations and two papers about purging malignant cells from bone marrow grafts. That is haematology, and a completely different sense of the word.

Dermatology does have a word for the period people are describing. It isretinization, and it means the redness, dryness, flaking and stinging that come with starting a retinoid. Notice what it does not mean: new spots.

Where the 28 days came from

The 1965 paper that produced it is free to read, and the sentence that matters is plain arithmetic. The authors measured how long cells took to travel through the living layers of the epidermis and got 13 days. For the outer dead layer they did not measure anything: they took 13 to 14 days from a different study, by Rothberg, and added the two together.

Verbatim from the paper: "When the turnover time of the normal stratum malpighii determined in this study (13 days) is added to the turnover time of the stratum corneum from the Rothberg study (13-14 days), a value of 26-27 days is found for the turnover time of the entire epidermis."

It is worth saying clearly that this is not sloppy science. Combining your own measurement with a published one was, and is, an ordinary thing to do, and the authors describe exactly what they did. The problem is what happened next: the sum became a fact, the fact became a round number, and the round number outlived its own correction.

Every figure ever published, and the one everybody quotes

In 1972 Halprin reviewed the whole literature on epidermal turnover in the British Journal of Dermatology and concluded that the journey probably takes 40 to 56 days. A measurement in 2017, using a completely different method, landed at 36.2 days. The 28 that consumer skincare repeats is lower than every estimate published since that 1965 sum was corrected.

pre-1956 Early mammalian studies, reviewed by Leblond and Walker
1961 Rothberg and colleagues
1965 Weinstein and van Scott
1972 Halprin, critical review of the literature
2017 Maeda, measured by UVA melanin clearance

The 2017 row is drawn as the study's stated 36.2 days plus and minus its standard deviation of 4.0. The 1965 row is the addition sum in the figure above. Every other row is the range as the source printed it.

Two cautions, because a chart flattens things a sentence would not. "Turnover time" is used for two different journeys, and mixing them is half the confusion in this subject. And the figure moves with the person: it lengthens with age, by more than ten days in older adults, and it is faster on the face than on covered skin.

Both are called "turnover time" in the literature. Almost every consumer article gives one number without saying which journey it means.

What actually gets measured, and when it peaks

Acne trials do not count purges. They score local tolerability at every visit: erythema, scaling, dryness, burning and stinging. That is retinization, and across independent trials and reviews it peaks almost immediately.

Each dot is the week a source reports irritation peaking on the face. Where a source gives a range, both ends are drawn. Nothing here is a magnitude: these are the reported peak weeks, not scores, because the sources report timing in words rather than a curve we could honestly redraw.

And here is the distinction that the whole subject turns on, which no chart can do for you.Irritation is not the same as breaking out. Everything above measures redness, flaking and stinging. What people call purging is new spots. The trials score the first and not the second, so none of this tells you when a breakout ends. It tells you that the discomfort people are enduring for six weeks is usually worst in the first one.

Does the purge itself happen?

This is where we expected to find the evidence and did not. A 2026 acne guideline advisory, written by dermatologists to guide prescribing, says it plainly:

"Flare-ups or transient worsening of acne have been reported after starting topical retinoids, such as tretinoin, although these occurrences arenot commonly documented in published literature."
Advisory statements on trifarotene, J Cosmet Dermatol, 2026

The initial worsening that is well documented belongs to a different drug: oral isotretinoin, a prescription systemic medicine. It appears across the literature, it has proposed mechanisms, and in at least one published case it was severe enough that treatment was stopped. The consumer story has borrowed a real phenomenon from a pill and attached it to a serum.

None of which means you are imagining your skin. It means nobody has published a study of what you are going through, and the confident six-week timetable you were handed came from arithmetic rather than observation.

So what should you actually expect

Honestly: less certainty than you were promised, and a shorter worst week than you were warned about.

The measured discomfort curve peaks in the first week or two on the face and eases as tolerance develops. If your skin is still getting worse in week six, the literature offers you no reason to believe that is a normal stage of a process, because no such process has been characterised. That is a reason to ask someone, not a reason to endure quietly.

And if a product told you to expect a purge, notice what that claim does. It converts the product making your skin worse into evidence that it is working, for a period defined by a number from 1965 that was corrected in 1972. That is a very convenient thing for a product to be able to say.

The one you actually started

Everything above is about the idea. If you want the answer for the specific thing now sitting on your bathroom shelf, each of these takes the same evidence and applies it to one ingredient — including where it sits on the labels of real products, which decides whether you are getting a working dose of it or a trace.

All of them, grouped by how much anybody actually knows.

How a correction gets lost

This is the part that is not really about skin. A number entered the literature as a convenience, was corrected seven years later in the field's leading journal, and the correction never travelled. Sixty years on, the pre-correction figure is the one on every product page, and the paper that fixed it is one search away.

The 1965 figure travelled. Both later corrections sit beside the path rather than on it. Nothing here is a metaphor: the dates, the numbers and the journals are in the sources below, and four of the six are free to open.

We are not claiming anyone acted in bad faith, and we did not audit the whole internet. We read one page of results, and every article on it repeated the same uncited number. What that suggests is ordinary and human: writers cite writers, the round number is easier to carry than the range, and nobody has an incentive to go back to 1972.

How we know this

Every figure above was read in the original paper, at the level stated. Where we could only reach an abstract, we say so and quote nothing beyond it.

Weinstein GD, van Scott EJ. Autoradiographic Analysis of Turnover Times of Normal and Psoriatic Epidermis. J Invest Dermatol 1965;45(4):257-262.Full text. Open Archive at the publisher, PDF only.
Halprin KM. Epidermal "turnover time" - a re-examination. Br J Dermatol 1972;86(1):14-19.Abstract. The full text is paywalled, so we quote the conclusion and not the reasoning.
Maeda K. New Method of Measurement of Epidermal Turnover in Humans. Cosmetics 2017;4(4):47.Full text, open access. Six healthy Japanese men, inner forearm.
Bridging Gaps in Malaysian Acne Vulgaris Guidelines: Advisory Statements on Trifarotene. J Cosmet Dermatol 2026.Full text, open access.
Phase 3 and phase 4 trifarotene trials, and a post-NICE acne review, for the reported peak week of local irritation.Full text, open access, via the Europe PMC full-text API.
Europe PMC, whole-database searches for "skin purging" and for purging with acne or retinoids.Run 2026-08-05. Anyone can repeat them.

We have not read Rothberg 1961, whose 13 to 14 days makes up half of the 1965 sum, or Baker and Kligman 1967, which originated the dye method behind the shorter figures. Both are paywalled. Neither changes what the papers we did read say about themselves.

Questions people actually ask

Is skin purging real?

The experience is real. Whether "purging" is a real category is a different question, and the honest answer is that medicine does not use the word. A search of the entire biomedical literature for "skin purging" returns two papers: a review of a Chinese medicinal plant and a report on forensic medicine from 1857. Dermatology's word for the adjustment period after starting a retinoid is retinization, and it describes redness, dryness, flaking and stinging rather than new spots.

How long does skin purging last?

Nobody has measured it. The "4 to 6 weeks" you will read everywhere is derived from a cell-turnover figure of 28 days, and that figure is a rounding of a 1965 arithmetic sum that the field revised upward to 40 to 56 days in 1972. What has been measured, repeatedly, is retinoid irritation, and it peaks in the first week on the face and declines from there.

Where does the 28-day skin cycle come from?

From a 1965 paper. Weinstein and van Scott measured cell transit through the living layers of the epidermis at 13 days, took a figure of 13 to 14 days for the outer layer from a different study by Rothberg, added the two, and wrote that "a value of 26-27 days is found for the turnover time of the entire epidermis". In 1972 Halprin reviewed the whole literature and concluded the real figure was probably 40 to 56 days.

Is purging different from breaking out?

That distinction is the thing everyone wants and nobody has evidence for. A 2026 acne guideline advisory states that flare-ups after starting topical retinoids "are not commonly documented in published literature". The well-documented initial worsening belongs to oral isotretinoin, a prescription systemic drug, not to an over-the-counter serum. So if a rule for telling the two apart exists, it has not been published.

Should I stop using my retinoid if my skin gets worse?

That is a question for a clinician, not a website, and especially if you were prescribed the product. What we can tell you is what the trials record: local irritation is usually mild to moderate, peaks early, and tolerance develops with continued use. In at least one published case, an initial worsening was severe enough that treatment was stopped. Both things happen and neither is a rule. Here is when to see a dermatologist.

Corrections and updates

8 August 2026Added per-ingredient answers. The essay explains where the 28-day clock came from; the four new pages answer "does mine do it", which measurement showed is a separate question with a separate results page.

Last re-read against its sources on 5 August 2026. If you find an error, tell us and we will fix it and say so.

Get the next one

We read the primary papers, write to the researchers, and print what they say. If that is useful, the free plain-words routine guide comes with it. No spam, and nothing sold here.