How we grade evidence

GRADE A·Established

Repeated human randomized trials showing outcomes that matter, not just biomarkers.

GRADE B·Promising

Human trials show real biomarker or functional effects; outcome and long-term data still missing.

GRADE C·Early

Animal or small pilot data only. No dependable human evidence yet.

GRADE D·Insufficient or unsafe

No credible human evidence, or the known risks outweigh what is known about the benefit.

Only research conducted in humans can set a grade. In practice that means randomized controlled trials, and where those do not exist, controlled human studies of a lower design.

Animal and cell studies never raise a grade. Most longevity claims in circulation come from mice, and the leap from a mouse lifespan result to a human recommendation is exactly where the field goes wrong. Where the only evidence is preclinical, we say so and grade it C.

We also record what is missing. Every compound page carries a section stating what has not been measured — trial length, disease outcomes, populations never studied — because absence of evidence is information a reader is entitled to.

A compound that moves a marker has not been shown to change a life. Raising NAD+, lowering a lipid, shifting an epigenetic clock — these are reasons to keep studying something, not proof it works. A grade of A requires outcomes that matter to a person: disease, function, survival. Almost nothing in longevity currently qualifies, and we would rather say that than inflate the grades.

A grade applies to a compound at the dose and in the population that was studied. A trial in adults with prediabetes does not license a claim for healthy thirty-year-olds, and a product dosed below the studied range does not inherit the trial's evidence. Where a commonly sold dose falls short of what was tested, the page says so.

They often do. Where two credible trials reach different conclusions, we state the disagreement rather than choosing the flattering one. A reader deciding whether to spend money and take a substance is better served by an honest conflict than a tidy answer.

A grade reflects the evidence on the date shown at the top of each page. New trials can raise or lower it, and a downgrade is as publishable as an upgrade. When a grade changes materially, the page records what changed and why.

It is not a recommendation, and it is not advice for you specifically. A well-evidenced compound can still be wrong for you because of a medication you take, a condition you have, or a pregnancy. That is what the interactions section on each page is for — and why the site keeps telling you to involve your own clinician.