How we grade the evidence
Every claim links to its source, every time-sensitive fact carries a date, and every compound is graded the same way.
Evidence tiers
- ATwo or more large human trials
- BAt least one controlled human trial
- COnly small or uncontrolled human studies
- DAnimal studies only
- ELab or anecdotal evidence only
A tier describes the best human evidence for any use of that exact molecule. A related molecule does not count: studies of full-length thymosin beta-4 say nothing direct about the TB-500 fragment. A high tier means the molecule has been tested in people. It does not mean a gray-market vial is safe or contains what its label says.
Does it work?
The tier asks whether a molecule has been tested in people at all. "Does it work?" asks a narrower question: for the use people mostly take it for, what do controlled human trials show?
- Benefit shown Controlled human trials of the main use found a clear benefit.
- Mixed results Trials disagree, are small, measured only lab markers, or studied a narrower group than the people who buy it.
- Untested No controlled human trial of the main use has public results.
- No clear benefit Controlled human trials of the main use found no clear benefit.
A drug can have a high tier and still be untested for the way it is sold. Tesamorelin, for example, is approved for belly fat in people with HIV, but is sold mostly for fat loss in people without it.
Legal status at a glance
Each compound page shows four cells (FDA, US pharmacies, Sport, Australia), each with its source. US pharmacies means whether a state-licensed compounding pharmacy may make it under FDA's rules for bulk ingredients. Sport means the World Anti-Doping Agency's Prohibited List; where a compound is not named but falls under a catch-all class, we say so and call it our reading. The legal tracker puts every compound side by side and logs what changed.
Doses
Each page keeps three layers apart. In human studies: doses given in completed trials or written on an approved label. In animal studies: doses as reported, with no conversion to a human dose. What users report: dose ranges from community or seller pages, marked anecdotal. None of these is a recommendation. We do not publish dose calculators, mixing or injection instructions, or advice on where to buy.
Popularity
The ranking on the home page uses test requests to Finnrick, a US testing service, from May 2025 to May 2026. It is a proxy: the people who pay for tests are more careful than most buyers, and free testing is US-only. The People Also Ask questions come from our own snapshots of Google results in US English: 24 peptide searches on 30 September 2026 and 10 skincare and collagen peptide searches plus 12 hair, GLP-1, safety and women's health peptide searches on 8 October 2026.
Sources
We prefer primary sources: trial papers, drug labels, trial registries and regulators' own pages. News reports are used for news events. Every factual sentence carries a numbered link to its source, and every source shows its publisher and date. A source with no date of its own, such as a database or a search, shows the date we read it, since its contents can change. Seller pages are cited only as evidence of how products are marketed, and are not linked.
Dates
"Last reviewed" is the date we last checked a page against its sources. Legal status changes quickly, so each page says when its status was checked, and the change log lists every update.