Cosmetic hair growth serums

Key takeaways

• No cosmetic active has evidence close to minoxidil or finasteride. The best-supported cosmetics reach fairly good evidence (adenosine) or limited evidence (caffeine and a few branded peptide blends).
• Almost all positive human data are small, short (12-24 weeks), run or paid for by the ingredient maker, and often test a blend, so no single ingredient can be credited.
• In the US, a product that claims to regrow hair is legally a drug claim, and no non-approved over-the-counter ingredient is recognised as safe and effective for it. "Cosmetic" serums therefore use vague wording such as "supports" or "stronger-looking".
• Peptide and "stem-cell" brands (Redensyl, Capixyl, Procapil, AnaGain) have limited evidence at best. Baicapil, stemoxydine and nanoxidil rest on manufacturer claims only.
• Aminexil and kopexil, the minoxidil look-alikes, have no human trials; in one rat study the aminexil product grew the least hair of three products.
• Serums make most sense as an add-on to a proven treatment, or for people who cannot or will not use drugs. They should not delay seeing a clinician for rapid or patchy loss.

The science

Cosmetic actives aim at four mechanisms: widening the follicle's blood supply or nutrient delivery (niacinamide, methyl nicotinate), signalling dermal papilla cells to prolong the growth phase (adenosine, caffeine, melatonin), blocking local androgen or inflammatory signals (many botanicals, in the lab), and "activating stem cells" (branded blends, mostly lab or maker data). Most mechanism data come from cell culture or mice, and transfer to human scalps is unproven. Delivery matters: caffeine reaches follicles within minutes from a shampoo-type formulation, but staying on the scalp long enough is a different question.

What the studies show

• A meta-analysis of 7 adenosine trials (466 people) found all positive but with very low to moderate certainty, and the pooled 0.75% lotion result at 6 months was not statistically significant. Szendzielorz 2025
• A systematic review of 9 caffeine trials (684 people) found all favoured caffeine, but certainty was medium in 3, low in 1 and very low in 5. Szendzielorz 2025
• Only 8 non-prescription agents had any investigator-blinded controlled trial in a review of 13 studies. Gupta 2025
• A 2025 review of 24 papers and technical files on branded actives such as Redensyl, Capixyl, Procapil, Baicapil and AnaGain found mostly small studies of combined formulas, often conflicted and rarely compared with standard therapy. Bikash 2025
• In a 28-day rat regrowth model of 32 animals, the Aminexil product showed the lowest hair-growth effect of the three products tested. Orasan 2016

Evidence ladder

What it means by hair profile

• Early pattern thinning (any sex): serums are a low-risk complement; expect modest thickness gains, not restored density.
• Sensitive or itchy scalp: avoid essential oils and stinging "rubefacients" such as methyl nicotinate, and patch-test.
• Coarse or very curly hair: leave-in serums can add build-up. Oils may help dryness (a cosmetic effect) but not growth.
• Alopecia areata or scarring alopecia: do not self-treat with cosmetics; see a dermatologist.

Myths and facts

• Myth: "stem-cell" or "peptide" serums regrow hair like minoxidil. Fact: no head-to-head trial shows this beyond small, unblinded or placebo-free comparisons.
• Myth: natural means safe. Fact: rosemary, garlic, lavender, peppermint and other oils cause contact dermatitis.
• Myth: a company trial proves it. Fact: manufacturer-funded trials of blends cannot isolate the active, so read the design.

Bottom line

If you try a serum, pick one with adenosine or caffeine at studied concentrations, give it at least 4-6 months, photograph your hair under identical lighting, and add it to proven treatment rather than replacing it. Compare the price per month: expensive branded blends show no advantage in the evidence.