We graded 152 peptides. 64 have strong human evidence.
What a reader learns from putting every peptide on the same scale.
Peptides are sold two ways. Pharmacies dispense the approved ones against a prescription, with a label that lists trials, doses and side effects. Online vendors ship the rest in vials marked "for research use only", with a forum thread in place of a label. The same word covers both, and the marketing leans on that. A seller of an unapproved compound borrows the credibility of semaglutide without borrowing its 17,000-patient outcome trial.
So we put all 152 on one scale. Each compound gets a letter for the use it is best known for. A means the use rests on randomised controlled trials, or an approval granted on them. B means some human evidence with a gap: a single trial, small numbers, surrogate endpoints, or an approval for a narrow indication while the popular use has none. C means animal and cell studies, or uncontrolled human series. D means nothing meaningful has been published for the claimed use. The letter ignores how famous the compound is and whether the FDA has approved it for something else.
The count
64 peptides reach A, 43 sit at B, 40 at C and 5 at D. Put differently, 107 of 152 have any controlled human evidence for their known use, and 45 have none. On the regulatory side, 77 are FDA approved, 22 are in clinical trials, 37 are not approved for anything and 16 are cosmetic ingredients outside drug regulation.
The two axes do not line up. 15 approved drugs land at B, usually because the approval is for a rare indication and the compound is now sold for something the trials never tested. And 25 unapproved compounds sit at C or D, which is where most of the grey-market trade lives.
The popular ones
The names that fill peptide forums grade like this: BPC-157 (C), CJC-1295 (B), Epitalon (C), GHK-Cu (C), Ipamorelin (C), Melanotan II (C), MOTS-c (C), Selank (B), Semax (B), TB-500 (C). None of them reaches A. Most rest on rodent work, and in several cases the rodent work comes from a single laboratory, which is the weakest kind of replication. On the evidence checklist that sits on every page, 69 compounds fail the "more than one research group" test.
This is not a claim that they do nothing. It is a statement about what has been measured. A compound with forty rat studies and one twelve-patient case series has an evidence base that cannot tell you whether it works in a person, or what it does to one over a year.
What the grade does not say
A does not mean safe, and D does not mean dangerous. Semaglutide is A and carries a boxed warning. Several D compounds are probably inert. The grade answers one question: has anyone checked, properly, whether this does what it is sold for? Safety is covered on each page separately, from the same cited studies.
The grade also moves. A weekly job pulls new PubMed papers for every compound and revises pages when a trial, an approval or a safety signal lands. Each page shows when it was last checked and what changed.
How to use the site
Search a name or a brand, pick a body system on the map, or filter by what a compound is sold for. Every page opens with the claim beside what the studies show, then the reasons for the grade, then the research in plain language with a numbered citation on every factual paragraph. Two compounds can be set side by side. Nothing here is medical advice, and there is no dosing anywhere on the site. The method is described in full on the method page.
Published 2026-10-27. Figures are computed from the live index at build time, so they match the pages.