A serum costs more than a cream with the same actives in it. It is worth knowing what the extra word buys, and the answer starts somewhere unexpected: in the regulations, it buys nothing, because the regulations have never heard of it.
The word is not a regulatory category
On 2026-09-08 we searched the four parts of US federal regulation that govern cosmetics — 21 CFR part 700 (general), part 701 (labelling), part 720 (ingredient filing) and part 740 (warning statements), about 111,000 characters in all — for the word "serum".
It appears zero times.
There is no definition of a serum, no minimum concentration of anything, no viscosity band, no requirement that it differ from a lotion, and no rule about what may be called one. Anything in a bottle with a dropper qualifies, and so does anything without one. The same is true of "essence", "ampoule" and "booster". This is not a loophole; cosmetics regulation in the United States works on ingredients and safety and label truthfulness, not on product taxonomy. But it means the format is a marketing word doing the work of a specification, and it is priced as though it were the other way round.
What the label can and cannot tell you
The ingredient list is regulated, and the rule is more permissive than most people reading one assume.
21 CFR 701.3(a) requires ingredients to be declared "in descending order of predominance". Then 701.3(f) offers an alternative: ingredients may be grouped, with those "present at a concentration greater than 1 percent, in descending order of predominance", followed by those "present at a concentration of not more than 1 percent, without respect to order of predominance", followed by colour additives in any order.
That second group is where cosmetic peptides live. The Ordinary's copper serum declares copper tripeptide-1 at 1%; a product labelled 10% SNAP-8 contains roughly 0.005% acetyl octapeptide-3, because the raw material is a dilute supplier solution — the arithmetic is on our peptides for skin page. Every common cosmetic peptide is at or under the 1% line.
So the position of a peptide in a serum's ingredient list is legally arbitrary. Reading one tells you a single fact: the peptide is present at 1 percent or less. Whether it is at 0.9% or 0.0005% is not disclosed, and the order does not encode it. The widely repeated advice to check that an active appears "high in the list" is sound for a humectant at 5% and meaningless for a peptide.
The literature has a name collision
Counting the evidence for serums runs into a problem that is worth stating plainly, because it explains why confident category claims exist without much behind them.
In PubMed, "serum" appears in 1,326,569 abstracts — the blood fraction, one of the most common words in laboratory medicine. Searching for the cosmetic product by name instead ("facial serum", "cosmetic serum", "topical serum", "skin serum") returns 155 records in total, of which 12 carry the randomized-controlled-trial tag.
Then those twelve have to be read. At least four are not about a cosmetic serum at all:
| Record caught by the filter | What it actually studied |
|---|---|
| Ustekinumab in psoriasis, 2020 | Interleukin subunits in blood serum |
| Rilonacept in systemic sclerosis, 2018 | An interleukin-1 trap, blood markers |
| Antihistamine premedication, 2008 | Bee-venom immunotherapy |
| Advanced non-small-cell lung cancer, 1995 | Prognostic factors from serum chemistry |
A control search run in the same session — moisturizers crossed with skin, same randomized filter — returned 202. The moisturizer format has roughly seventeen times the randomized literature of the serum format under its own name, for a product type that costs less.
The biggest serum trial has been retracted
The largest randomized study of facial serums ran 130 participants for 24 weeks, double-blind, and compared two serums that differed by a liposomal delivery blend. Published in Skin Research and Technology in November 2023 by the research arm of a compounding-pharmacy supplier, it is the one study in this literature that isolated the thing a serum page should care about: the vehicle, with the actives held constant.
It was retracted in January 2026 (retraction notice, PMID 41574404). The notice gives three reasons in its own words: the article "was accepted for publication following an insufficient peer review process that does not meet the standards of the journal's peer review policy"; "post-publication review revealed that the study lacked the formal ethical approval required by U.S. regulations", because 45 CFR 46 does not exempt non-therapeutic cosmetic studies from institutional review; and "the original study data cannot be made available for independent verification due to closure of the laboratory where the research was conducted." The notice also records that the sponsor "emphasizes that the research was conducted ethically and without any misconduct or deception", and states that the retraction "does not constitute a finding of research misconduct."
Both halves matter. A retraction on process and ethics review is not a finding that the serums did nothing. It is a finding that the record cannot be checked — which, for the only trial that asked the format question directly, leaves the question unanswered rather than answered badly.
What is left, read honestly
The peptide serum study most often cited now is Peptide-pro complex serum: investigating effects on aged skin (PMID 35426243). Its authors are staff of Colgate-Palmolive's Research and Development, Dermal Clinical Research and Scientific Affairs groups, which is disclosed and ordinary in this field. The design is the part to read: the ex vivo portion compared the serum against a marketed competitor and an untreated control on skin explants, while the twelve-week clinical portion was monadic — "subjects functioned as their own control; evaluating change from baseline."
A monadic study cannot separate the product from twelve weeks of consistent twice-daily application, from the season, or from the moisturising base it is carried in. It can establish tolerability, which it did, and it can measure change, which it did. It cannot attribute that change.
The rest of the small set runs similarly: single-centre, sponsor-linked, often without a vehicle arm. Our page on peptide moisturizers found the same shape in a larger literature, where the one clean head-to-head against a control base found no significant difference on any endpoint while its own conclusion read as positive.
What the format does change
Two real differences, neither of them measured for a peptide.
A serum is typically water-based and low in viscosity, applied to bare skin before anything occlusive. A cream is a heavier emulsion. Occlusion raises hydration of the outer layer, which can raise permeability; a thicker emulsion also holds a water-loving molecule in its own oil-and-water film rather than releasing it. Those effects point in opposite directions and neither has been quantified for a cosmetic peptide in a finished product.
Packaging is the second, and it is more concrete. GHK-Cu degrades fastest under oxidative conditions, which is an argument for an airless pump or an opaque bottle over a wide-mouthed jar — a point about the container rather than about the word on it. Our copper peptides page covers the stability question and the mixing rules brands disagree on.
The honest summary
"Serum" tells you a texture and a price. It is absent from the regulations, it guarantees no concentration, and the ingredient list cannot resolve the concentration either, because everything below 1 percent may be listed in any order at all. The category's own literature is eight or so randomized trials, its largest has been withdrawn from the record, and the most-quoted peptide result comes from a manufacturer's study without a control group.
None of that means a peptide serum does nothing. It means the format is not the reason it would.
Sources
- 21 CFR 701.3, Designation of ingredients — eCFR, text retrieved 2026-09-08; paragraphs (a) and (f)(1)–(3) quoted.
- 21 CFR parts 700, 701, 720 and 740 searched in full for the term "serum" on 2026-09-08: zero occurrences.
- PubMed E-utilities counts run 2026-09-08, with a moisturizer control search in the same session; the twelve randomized records were read individually.
- RETRACTION: A Randomized, Double-Blind, Controlled Study Evaluating the Effects of Two Facial Serums on Skin Aging. Skin Res Technol 2026;32(1):e70333. PMID 41574404, retracting Skin Res Technol 2023;29(11):e13522.
- Moy M, Diaz I, Lesniak E, Giancola G. Peptide-pro complex serum: Investigating effects on aged skin. J Cosmet Dermatol 2023;22(1):267–274. PMID 35426243. Authors at Colgate-Palmolive Company.
Published by Peptide Glow Journal Editorial. No product is ranked, scored or affiliate-linked on this page.