Peptides for hair growth: what the studies show

The short answer

No peptide is an approved treatment for hair loss anywhere. A handful of small randomised trials did find a benefit in pattern hair loss, the largest in forty-five men, and none has been replicated. None appears in the systematic reviews that rank hair treatments, and the biggest human studies have no control group.

A peptide is a short chain of amino acids, the same building blocks that make up proteins, just fewer of them strung together. Peptides carry signals between cells and some of those signals are growth signals, so a peptide that tells a shrinking hair follicle to grow again is a reasonable hypothesis rather than an absurd one.

Androgenetic alopecia, the clinical name for ordinary pattern hair loss, is a condition in which hair follicles progressively miniaturise, meaning they shrink and produce thinner hairs, while the number of active follicles falls. It is heavily genetic: genome-wide studies, which scan a person's whole DNA for stretches linked to a trait, have connected more than 380 genomic regions to it [1].

A 2025 review of the condition in Nature Reviews Disease Primers sets out what actually works, and the list is short: transplanting a person's own androgen-resistant follicles from the back of the scalp, oral finasteride, and topical minoxidil [1]. No peptide is on that list. Peptides do appear in the same review, one sentence later, under emerging therapies, alongside messenger RNA packaged in liposomes and the bioengineering of entirely new follicles [1].

That does not mean nothing has been tested. Randomised, blinded trials of peptide products in pattern hair loss have been published and have reported benefits, in 22 Japanese men [15], forty-five Korean men [16] and 32 Thai men and women [17]. Every one is small, none has been repeated, and none reaches the reviews below. Getting those facts to sit together is most of what this article does.

Four different things are sold as hair peptides

The word covers four unrelated products: copper peptide serums sold as cosmetics, research chemicals such as PTD-DBM sold for experimental use, injectable growth factor cocktails given in clinics, and biomimetic peptide blends formulated into over the counter scalp products. They share a word, not an evidence base, and each has to be judged separately.

Almost every argument about this goes wrong in the first sentence, because the two people are talking about different products.

The most common by far is the copper peptide serum. Its active ingredient is usually GHK-Cu, a three amino acid chain (glycine, histidine and lysine) bound to a copper ion, sold as copper tripeptide-1. It makes no medical claim and is regulated as a cosmetic rather than a drug. Our guide to a GHK-Cu routine covers what it does on facial skin, a better documented question than what it does on a scalp.

Next is the research chemical, currently PTD-DBM: a powder sold by peptide vendors to mix at home, labelled not for human use while the marketing around it is about human hair. It sits in the grey space our article on vetting research peptides deals with. Then come the injectable cocktail given in a clinic and the biomimetic blend formulated into a retail product alongside plant extracts. In both the peptide is one line on an ingredient list with ten other lines, which matters more than it sounds: when a study of such a product works, nothing tells you which line did it.

Where the copper peptide claim comes from

It traces to a single 1993 paper. A copper binding peptide called PC1031 enlarged hair follicles on the back skin of fuzzy rats, and the authors described the effect as similar to topical minoxidil. That comparison is a rat comparison, on body skin rather than scalp, and it is the origin of a claim now printed on serum bottles.

Vendor pages for copper peptide hair products tend to say some version of "comparable to minoxidil in studies". Follow that sentence back far enough and it arrives at one paper, published in the Journal of Investigative Dermatology in 1993 by Uno and Kurata [4].

What that paper actually reports about a copper peptide is one sentence. A copper binding peptide designated PC1031 produced follicular enlargement on the back skin of fuzzy rats, covering the vellus follicles, and the authors wrote that the effect was similar to that of topical minoxidil [4]. Vellus hairs are the fine, barely pigmented hairs covering most of the body; terminal hairs are the thick pigmented ones on a scalp. Turning the first into the second is what any hair drug tries to do, so the result is interesting. It is also a rodent result, on body skin, and the abstract names no comparison group.

The rest of that paper is where the confusion comes from, because it does contain impressive primate work. In macaques with pattern baldness, topical minoxidil and diazoxide both induced significant hair regrowth on bald scalps, and a steroid 5 alpha-reductase inhibitor called 4MA, applied to non-bald preadolescent macaques, prevented baldness while controls developed it over 2 years of treatment [4]. None of those macaque results involved the peptide. A claim travelling from "copper peptide enlarged follicles in fuzzy rats" to "copper peptides work about as well as minoxidil" has borrowed the primate data next door.

Three decades later, nothing has replaced that sentence with a randomised trial of a copper peptide serum. The nearest thing is a 2016 Korean trial of a GHK complex delivered as a scalp spray [16], covered further down, and it is not the study the serum labels are pointing at.

The reviews that rank hair treatments list no peptide

Two recent network meta-analyses pooled the randomised trials and ranked treatments by how much hair density they add. One went hunting for over the counter alternatives and found seven with trial evidence behind them. The other searched four databases including the trial registry. Neither list contains a peptide, and the eligibility rules explain most of why.

A network meta-analysis is a statistical method for ranking treatments that have never been tested against each other directly, by chaining together the comparisons that do exist. It is the closest thing this field has to a scoreboard.

The first, in the Journal of Cosmetic Dermatology in 2025, compared conventional treatments against alternative, non-conventional, over the counter products. Its authors searched PubMed and Scopus without date restrictions on April 30, 2025, measuring everything on the change in total hair density at 24 weeks from baseline, in hairs per square centimetre [2]. They found 24 eligible trials covering eight conventional treatments and seven alternatives: topical ketoconazole 2%, a marine complex, topical procyanidin 0.7%, topical rosemary, topical melatonin, topical saw palmetto and topical watercress 2% [2]. No peptide is among them. The most effective comparator was oral dutasteride 0.5 mg, top of the ranking in both the base analysis (SUCRA 95.8%) and the severity-adjusted one (94%) [2].

The second, in Medicina in 2026, went wider, searching PubMed, Scopus, Web of Science and ClinicalTrials.gov on 23 January 2026 for off-label therapies. It identified 23 studies (15 of off-label treatments and 8 of topical minoxidil) and ranked topical minoxidil 5% most effective, followed by topical melatonin, topical diaminopyrimidine oxide 1%, topical procyanidin 1%, topical saw palmetto, a subcutaneous exosome formulation, topical cetirizine 1% and topical finasteride [3]. Again, no peptide.

Both absences are real, and the obvious explanation for them is wrong. The 2025 review required a study to test a conventional monotherapy or an over the counter product on the 24-week change in total hair density, and its authors deliberately picked 10 or fewer alternative agents, guided by earlier published reviews [2]. That is a narrow gate. The randomised peptide trials that do exist report a thick hair ratio at 4 months [15], a hair count inside a circle 1 cm across at 6 months [16] and a percentage change in terminal hairs [17], and every one of them tests a combination or a single small cohort. None lands cleanly in that network. The absence is about shape and selection, not existence, which is its own comment on how thin the evidence is.

The delivery problem comes before the evidence problem

Skin is built to keep molecules out. A long standing rule of thumb in dermatology holds that a compound has to weigh under 500 Dalton to be absorbed through it at all. Most cosmetic peptides are heavier than that and water loving, which is the wrong combination for crossing an oily barrier.

Before asking whether a peptide grows hair, it is worth asking whether it gets near a follicle. A follicle's growth machinery sits in the dermal papilla, a cluster of cells at its base, several millimetres down. A serum on a scalp has to cross the stratum corneum, the outermost layer of dead, tightly packed, fat-rich cells whose job is keeping things out.

In 2000, Bos and Meinardi proposed what became known as the 500 Dalton rule: that a compound's molecular weight has to be under 500 Dalton to allow skin absorption, and that larger molecules cannot pass the corneal layer [5]. A Dalton is a unit of molecular mass, roughly the mass of one hydrogen atom. Their argument was that common contact allergens, topical dermatology drugs and transdermal delivery drugs are all under 500 Dalton [5]. A rule of thumb rather than a law, but it sets the scale. A 2025 review of skin permeability for cosmetic peptides puts the consequence without hedging: peptides are not inherently designed to penetrate the skin, and their high molecular weight and hydrophilicity, meaning their tendency to dissolve in water rather than fat, pose significant barriers to transdermal delivery [6].

The exceptions matter, though. GHK-Cu is small, and the GHK complex in the randomised spray trial below has a molecular weight of 453.23, under the threshold [16]. A follicular duct is itself a hole in the barrier. But the rule explains why a peptide that impresses cells in a dish does nothing measurable on a head, and why the clinic products below are injected rather than rubbed on.

The randomised trials nobody cites

Four exist, and the three that report a sample size are all small. A Japanese lotion trial randomised 22 men to the same lotion with or without a five amino acid peptide. A Korean spray trial randomised forty-five men to a GHK complex or placebo. A Thai trial put a peptide blend against minoxidil in 32 people.

The randomised literature here is small and almost never cited by the people selling these products. The three trials that report a size enrolled 22 [15], forty-five [16] and 32 people [17].

The cleanest, published in 2016 [15], is the only study in this article that isolates a peptide. The compound is a five amino acid chain written Gly-Pro-Ile-Gly-Ser, which the same group had shown promotes proliferation of mouse hair keratinocytes and accelerates hair growth in mice [15]. It gave 22 Japanese men with androgenetic alopecia the same lotion with or without the peptide for 4 months, double-blind and randomised [15]. Because the only difference between the two bottles is the peptide [15], whatever the trial found belongs to the peptide and not to a vehicle or a plant extract. The authors concluded that it increases the thick hair ratio in living people [15]. The abstract is where the detail stops.

The one that matters for copper peptide claims, also 2016 [16], was a randomised, double-blind, six month study at Kyungpook National University Hospital. It gave forty-five men a nightly scalp spray of ALAVAX, a complex of 5-aminolevulinic acid and glycyl-histidyl-lysine (the copper-free form of GHK), at one of two strengths, or a placebo, with investigators and patients blinded until the study finished [16]. Hair count inside a circle 1 cm across rose by 52.6 in the stronger group and 71.5 in the weaker one, against 9.6 on placebo [16].

Those numbers are easy to over-read. The significance marks are within-group change from baseline, not a comparison against placebo; the only significant between-group result at 6 months was the ratio of change in the weaker group against placebo [16]. The dose response runs backwards. The standard deviations are roughly as large as the effects, 45.7, 44.9 and 45.1 [16], and hair length and thickness did not differ between groups [16]. ALAVAX is a two-part molecule, so nothing separates the peptide from the 5-aminolevulinic acid, and the authors call it one of the complementary agents rather than a treatment [16].

Two more sit at the edges. A 2020 Thai trial randomised 32 men and women with mild to moderate androgenetic alopecia, triple-blinded, to twice-daily applications of either a herbal combination containing acetyl tetrapeptide-3 or a 3% minoxidil solution for 24 weeks, and found no statistically significant difference in the increase in terminal hair counts, 8.3% against 8.7% [17]. That is the closest thing to a peptide product tested head to head against minoxidil, and it has no placebo arm, three active ingredients and thirty-two people; its authors ask for larger groups [17]. A 2018 paper on water-soluble chicken egg yolk peptides reports that they stimulate VEGF production and human dermal papilla cell growth and improve hair growth in female pattern hair loss [18], though its abstract gives no trial design.

The other human studies, and what they can carry

The rest of the human research cannot isolate a peptide. The largest gave 1000 patients a six ingredient injection with no control group. One randomised trial studied alopecia areata, a different disease. One cosmetic study of thirty people tested a peptide blended with red clover extract. Every one of them is a mixture, an uncontrolled design, or both.

The rest of the human literature is quoted far more often than any of those trials and carries far less.

The largest is a 2018 pilot study of an intradermal injection formula given to 1000 patients once every 3 weeks for eight sessions [7]. The results sound strong: hair fall fell significantly in 83% of patients on the hair pull test, and at 1 year total hair count rose significantly [7]. Two things decide how much that carries. It was open-label, prospective and single-arm, with no comparison group at all [7]. And the formulation contains vascular endothelial growth factor, basic fibroblast growth factor, insulin-like growth factor, keratinocyte growth factor, thymosin beta-4 and copper tripeptide-1 [7]. Two of those six are peptides, and nothing in the result is attributable to either. A 2025 hair transplant trial of a related injectable, QR678 Neo, in 112 males aged 25 to 50 did have a comparison group, which reached a higher global photographic assessment score (8.87) and terminal hair count (181.02) than conventional transplant alone [8]. Better design, still a mixture.

A randomised trial of a biomimetic peptide product is also cited here often, and it studied alopecia areata [9], an autoimmune disease in which hair falls out in patches. Its results do not transfer to pattern hair loss [9].

Finally, a 2013 cosmetic study tested a biomimetic peptide together with a Trifolium pratense, or red clover, flower extract [10]. Thirty volunteers with recessing hair took part in a randomised, placebo-controlled design measured by TrichoScan, an automated photographic hair count [10]. Within 4 months, anagen hair (hair in its active growing phase) rose by an average of 13% and telogen hair density (hair on its way out) fell by 29%, against a placebo group that moved the other way on both [10]. Thirty people testing a mixture in which the plant extract is at least as plausible a candidate as the peptide.

PTD-DBM and thymosin beta-4: a target and an animal

Both are real biology with no human trial behind them. The CXXC5 work found a brake on a growth pathway that is more active in balding scalp, and showed that a competing peptide released it. Thymosin beta-4 grows hair in mice and cashmere goats. Neither has produced a published human trial result.

PTD-DBM is the peptide sold hardest to people who read about hair loss online, and the science it comes from is legitimately good. In 2017, a Korean group reported that a protein called CXXC5 acts as a brake on the Wnt/beta-catenin pathway, one of the core signalling systems that tells a follicle to enter its growth phase [11]. They found CXXC5 upregulated in miniaturised hair follicles and arrector pili muscles in human balding scalps, which is a genuine human finding, then showed that disrupting the interaction between CXXC5 and a partner protein called Dishevelled with a competitor peptide activated the pathway and accelerated both hair regrowth and the formation of brand new follicles in healing wounds [11].

Read the paper's own conclusion and the gap becomes obvious. The authors write that the CXXC5-Dishevelled interaction is a potential target for the treatment of hair loss [11]. A target is not a treatment. The step between them is a clinical trial. Searching the United States trial registry for PTD-DBM or CXXC5 returns no hair loss study, and no human results have been published since. Every claim in a vendor listing extrapolates from animal and cell work.

Thymosin beta-4 is the same shape of story, further along in animals and no further along in people. It is a small protein that binds actin, part of the internal scaffolding of cells, and a 2021 review sets out its role in hair follicles [12]. The body's own thymosin beta-4 activates the mouse hair follicle cycle; given from outside, it increases hair growth in mice and cashmere production in cashmere goats by increasing the number of secondary follicles [12]. Mice, rats and cashmere goats are the entire dataset, and even there the review notes that the molecular mechanisms have rarely been reported [12]. TB-500 is a fragment of this molecule, and the leap from a goat's fleece to a human hairline is not small.

What is actually in the pipeline

One randomised trial is now running. It started in May 2026 at a dermatology institute in Thailand, compares a peptide serum against 2% minoxidil in 80 people over 24 weeks, and is masked three ways. That makes it larger than any randomised peptide trial above that reports its size.

The preclinical work has not stopped, and some of it is careful. A 2026 paper screened over 200 human protein-derived peptides for their ability to cross cell membranes, picked one, named it DualPep-ALO, and tested it in ex vivo human scalp tissue, real scalp skin kept alive in a dish after surgery [13]. There it lengthened follicles, held them in their growth phase and raised several growth factors, which the authors describe as outcomes comparable to minoxidil [13]. Their conclusion asks for what is missing: clinical investigation to validate efficacy and safety [13].

Which brings us to the genuinely new thing. A randomised trial registered as NCT07536100 is recruiting now at the Institute of Dermatology in Thailand, with a cosmeceutical company as collaborator [14]. It assigns participants in parallel groups to either a peptide factor hair serum or a 2% minoxidil solution for 24 weeks, with an estimated enrolment of 80 and masking of the participant, the investigator and the outcomes assessor [14]. It started on 2026-05-05 and its primary outcome is the change from baseline in hair count within roughly a 1 square centimetre target area at the vertex, by standardised macrophotography at weeks 12 and 24 [14].

That design is what this field has been short of: randomised, masked three ways, against an active comparator known to work, and larger than any of the randomised peptide trials above. Until it reports, the defensible position is the uncomfortable one: a few small positive trials, none replicated, a great deal of animal work, products sold years ahead of the data, and the treatments with real evidence behind them still the ones dermatology already had.