Peptides for Hair Growth: What the Evidence Actually Shows

The short answer, before the shopping

Search this phrase and almost everything that comes back is a shop. Here is the published record instead, ranked the only way that means anything: by what was measured, in what, and against what control. As of September 2026 no peptide has an approved indication for pattern hair loss anywhere. The approved options remain topical minoxidil at 2 and 5 percent and oral finasteride at 1 milligram, approved in December 1997 for men only.

Four peptides have randomised human trials with a hair endpoint. Three are still sold for hair, and each of those trials carries a catch that belongs in the same sentence as the result; the fourth was tested properly and failed. A tier below sits human tissue in a dish. Below that, rodents. Below that sits the group people actually buy and inject, where the finding is not a weak result but the absence of any result.

None of that makes the category fraudulent. Some of the biology is genuinely interesting and nothing on the record suggests bad faith. It makes the field early, and the distance between how early it is and how confidently it is sold is the thing worth knowing.

The ranking, sorted by what was measured

The tiers below do not rank how promising a mechanism sounds. They rank the kind of experiment that has actually been run. A randomised trial in people is a different object from a follicle in organ culture, which is different again from a depilated mouse, and no amount of mechanistic elegance promotes a compound between rows.

Hold the approved drugs alongside as a yardstick. The two pivotal vertex trials of finasteride randomised 1,553 men and reported a mean difference of 107 hairs per 5.1 square centimetres at twelve months and 138 at twenty-four. Topical minoxidil at 5 percent beat both 2 percent and placebo on target-area hair count in 393 men over 48 weeks. Nothing in the peptide column is within sight of that.

The cautionary case sits inside the table. FOL-005, an osteopontin-derived peptide, is the only single peptide ever put through a properly powered vehicle-controlled scalp trial: 209 men, sixteen weeks, 6.6 hairs per square centimetre in the top-dose arm against 5.6 on vehicle, p = 0.83 between groups. Hair growth rose significantly from baseline in that arm and still did not beat vehicle. The programme was discontinued.

What a hair peptide would actually have to change

Each scalp follicle cycles through anagen, active growth lasting typically two to seven years; catagen, a regression of about two to three weeks; and telogen, roughly a hundred days of rest, followed by exogen, the release of the club hair. Staging 3,607 occipital follicles from fourteen men put 88.99 percent in anagen at any moment. Around ten to twelve percent are not growing.

In pattern hair loss, dihydrotestosterone generated by 5-alpha-reductase acts on androgen receptors in the dermal papillae of susceptible follicles. Over successive cycles anagen shortens, the empty interval between them lengthens and the dermal papilla shrinks. Thick pigmented terminal hairs above 60 micrometres are replaced by fine ones below 30. The terminal-to-vellus ratio falls from about 36 to 1 toward 3 to 1, and shaft diameter from 82.7 micrometres to 61.1.

That is the target. A hair peptide has to lengthen anagen or enlarge the dermal papilla, and none of the peptides marketed for hair touches androgen metabolism. One aside worth carrying: the familiar fifty to a hundred hairs a day is arithmetic back-calculated from an old telogen percentage, not a measurement. Standardised wash tests give 27.9 hairs in unaffected people and 52.2 in pattern loss.

The GHK trial that was not a GHK-Cu trial

The strongest-looking result in the copper-peptide literature is a six-month randomised, double-blind, placebo-controlled trial in 45 Korean men with Norwood-Hamilton stage II to V loss, three arms of fifteen, a spray applied nightly with no light activation. Hair count inside a one-centimetre circle rose by 52.6 at the higher concentration and 71.5 at the lower, against 9.6 on placebo.

Then the catch. The spray was ALAVAX, a single covalent conjugate of 5-aminolevulinic acid with glycyl-histidyl-lysine, and the peptide in it was copper-free. With no GHK-alone arm and no 5-aminolevulinic-acid-alone arm, the trial cannot say which half did anything. Only the lower concentration beat placebo between groups, as a ratio of change, 2.38 against 1.21; the higher did not, which is an inverted dose-response. Hair length and thickness did not differ at all, and the placebo arm started with the highest baseline count.

So this is evidence about ALAVAX, generating a hypothesis about GHK, in about fifteen men per arm, at one centre, never replicated in the decade since. It is not evidence for GHK-Cu, a different molecule carrying a copper ion. And there is a second gap the sellers never mention: nobody has tested injected GHK-Cu for scalp hair at all.

Topical or injected, and the comparison nobody has run

Injection into the dermis bypasses the stratum corneum outright, so as a matter of physics it delivers more peptide to the tissue than a cream does. The case for topical is not superior delivery. It is convenience, cost, daily self-administration and avoiding a procedure, plus the fact that the follicular opening is a real second route and the dermal papilla sits a few millimetres down inside it.

One argument should be retired. The 500-dalton rule is a viewpoint article from the year 2000 with no permeation data of its own, built from the observation that contact allergens and topical drugs happen to be small. It should not be aimed at GHK-Cu in any case: the free tripeptide is about 340 daltons and the copper complex about 402. Size is not the barrier. Hydrophilicity and charge are.

What does not exist is a comparison. No study in any species has tested the injected and topical routes against each other for any hair peptide. The injected human data that do exist are uncontrolled series of cocktails in which copper tripeptide-1 is one of six to twenty actives. That is not a finding that injection fails. It is a finding that the question was never asked.

Acetyl tetrapeptide-3 and GPIGS: a four-active tonic and a maker-run trial

The second randomised trial is the one behind acetyl tetrapeptide-3. Over 24 weeks, 32 Thai adults with mild-to-moderate pattern loss, mean age 41, were randomised triple-blind to a scalp tonic containing acetyl tetrapeptide-3, biochanin A from red clover, Panax ginseng extract and sage oil, or to 3 percent minoxidil. Terminal hair count rose 8.3 percent on the tonic and 8.7 percent on minoxidil, p = 0.306 between arms at week 24.

The catches compound. Four actives, so nothing in that 8.3 percent belongs to the peptide rather than to the 5-alpha-reductase-inhibiting isoflavone beside it. No placebo or vehicle arm, so neither gain can be separated from what an alcohol-and-panthenol vehicle applied twice daily would have done. And with seventeen subjects against fifteen and no equivalence margin, a non-significant difference is a failure to detect one, not a match.

The third trial is the cleanest in design and the most compromised in provenance. GPIGS, a bacterially derived pentapeptide sold as Pentapeptide-20, was tested in a lotion against the identical lotion without it, in 22 Japanese men over four months. The single difference between arms was the peptide, which no other trial of a marketed hair peptide can say. The proportion of hairs at least 60 micrometres thick rose more than on vehicle, p = 0.006, and blinded photographic grading improved, p = 0.020.

And the catch. Hair density, the conventional measure of whether there is more hair, did not rise significantly, so what shifted was the thickness distribution of hairs already present. The manufacturer ran and published it, with twenty-two men at one centre, and nobody has repeated it in the decade since. Its European cosmetic listing assigns it the function hair conditioning.

Human tissue is not a human being

A tier below the trials sits work on human tissue. AHK-Cu, the copper complex of alanyl-histidyl-lysine, was tested in 2007 on follicles from ten volunteers kept in organ culture for twelve days. Elongation increased significantly against vehicle across the picomolar-to-nanomolar range, and cultured dermal papilla cells proliferated. That is a real effect in real human tissue.

It should also not be oversold. The response is biphasic and inverts: one log unit higher the same compound inhibited elongation by 14.8 percent, and two log units higher the inhibition was close to total. The only direct measurement of apoptosis was explicitly not significant. It is one paper from one laboratory, never replicated in nineteen years, with the compound supplied by a copper-peptide company. And AHK-Cu is not copper tripeptide-1: the line circulating about AHK-Cu injections retaining hair around the injection site is a mouse result belonging to a different, multi-ingredient product.

The newest entry sits in the same tier. A 2026 paper describes a synthetic peptide built from the Helix C-1 region of erythropoietin, which engages the erythropoietin receptor on dermal papilla cells, raises IGF-1 secretion, lengthens hair shafts in human follicle organ culture and accelerates anagen entry in depilated mice. No person has received it and no trial is registered. The claims circulating that it triples IGF-1 or matches minoxidil trace to press coverage, not to the paper. An isolated follicle is denervated, unvascularised and heading for catagen within about two weeks.

Rodents: thymosin beta-4 and PTD-DBM

Thymosin beta-4 has the older rodent story. Applied topically to shaved rat skin it roughly doubled the number of growing follicles against vehicle after seven days. Mice engineered to over-express it regrew depilated hair in eleven days against thirteen for wild type, while knockouts took sixteen, and VEGF in that skin rose 2.33-fold at the message level and 3.17-fold at the protein level. The groups were three animals each.

The single experiment ever done on human tissue points the other way. In serum-free organ culture of human scalp follicles, thymosin beta-4 at 1,000 nanograms per millilitre for seven days reduced hair-shaft production by roughly ten to twenty percent and induced catagen prematurely. It is the only human-tissue data point in the file, and it is negative.

PTD-DBM comes from better-motivated biology. CXXC5 is a Wnt-target feedback brake binding the Dishevelled PDZ domain, and it is over-expressed in the miniaturised follicles and arrector pili muscles of balding scalp, which is an association rather than a demonstrated cause. The peptide competes for that domain, and in mice it reactivated Wnt signalling and accelerated regrowth and wound-induced follicle neogenesis, with the largest effect when valproic acid was co-applied. Neogenesis of that kind barely happens in adult human scalp. There is no human trial and no human tissue study, the senior author heads the company licensing the peptide, and the laboratory itself moved on to small molecules.

The four injectables nobody has ever tested on hair

This is the most useful fact in the article, so it gets stated without hedging. No study in any species has tested BPC-157, CJC-1295, ipamorelin or MOTS-c for hair growth or hair loss. Not a trial, not a mouse, not a follicle assay. A registry search across all four returns seventeen trials between them, covering hamstring strain, rotator cuff tears, post-operative ileus, gastrointestinal dysmotility, HIV-associated visceral obesity and prediabetes. Not one carries a hair, scalp or alopecia endpoint.

The near-misses matter, because they are where the marketing copy comes from. The MOTS-c papers about hair cells concern cochlear sensory hair cells, not follicles. The one genuine follicle observation in this group is a line in a 2001 mouse burn study noting more follicles preserved in skin treated with a BPC-157 cream, which is wound healing. And the thirty-patient BPC-157 alopecia study quoted across supplier pages has no citation, no indexed record and no registration.

The growth-hormone rationale behind the secretagogues fails at a specific link. IGF-1 genuinely does prolong anagen in cultured human follicles, but only in insulin-free medium; alongside insulin it did nothing. Growth hormone itself had no effect in that same experiment, and in a later one recombinant human growth hormone pushed female human scalp follicles into premature catagen through TGF-beta2. Follicle IGF-1 is a local paracrine signal, and nobody has shown that raising the circulating concentration raises the one inside the follicle.

The natural experiments are equally unhelpful to the pitch. After surgery corrected their growth-hormone excess, 263 of 484 patients with acromegaly reported scalp hair loss within three to six months, against 6 of 167 after other pituitary surgery. But lifelong deficiency of growth-hormone-releasing hormone in one described cohort came with youthful hair and no baldness, which is the specific rebuttal to a GHRH analogue sold for hair.

TB-500 does not inherit thymosin beta-4's evidence

This one deserves separating out, because the evidence appears to transfer and does not. Two independent doping-control laboratories, working from separately sourced product, identified the active ingredient of TB-500 as an acetylated seven-residue peptide corresponding to residues 17 to 23 of thymosin beta-4. That fragment weighs about 889 daltons; the parent is 43 residues and about 4,963, a difference of roughly 5.6-fold.

Every published hair study of thymosin beta-4 used the full-length protein or altered the endogenous gene in mice. No study in any species has put the seven-residue fragment on a hair follicle. The anti-doping list treats TB-500 as a derivative rather than the same substance, which is the correct relationship. The position is not that TB-500 fails for hair; it is that the experiment has never been run.

There is a reason not to assume the fragment behaves like the parent. In serum and in rats it is cut further to shorter peptides, and in one wound-healing assay the activity tracked a metabolite rather than the parent compound. Thymosin beta-4's own N-terminal tetrapeptide is a separate bioactive entity with its own pharmacology, which is the in-house demonstration that fragments of this protein are not miniature versions of it. TB-500 is also a trade name rather than a specification.

Reading the claim in front of you, and what the evidence supports

The failure modes repeat, which makes them checkable. Was the peptide tested on its own or inside a blend? Was the trial run by the company selling the ingredient? Were the subjects people, or follicles in a dish? Was the endpoint a hair count in a marked patch of scalp, or a photograph and a satisfaction score? And is the compound on the label even the molecule that was in the study?

One regulatory point sits underneath all of it. A cosmetic may describe appearance. A product that claims to regrow hair has made a drug claim, and in the United States that makes it an unapproved new drug whatever is in the bottle. The cosmetic registry lists the best-tested peptide here under hair conditioning.

So, honestly: the approved options have vastly better data, on the endpoints regulators accepted, in samples an order of magnitude larger. The peptide that earns the most attention is GPIGS, because it is the only marketed peptide tested against its own vehicle with the peptide as the sole difference, and its catch is that the manufacturer ran and published it and nobody has repeated it.

And the route matters more than the ingredient lists suggest. Every randomised human trial in this field was topical. Injected GHK-Cu has never been tested for scalp hair by anyone, and the four peptides most often injected for hair have never been tested for hair in any species. No amount of before-and-after photography changes that.