How we grade the evidence
Every page starts with a verdict and a grade. The grade tells you how strong the research behind the main claim is, not how confident anyone sounds.
The four grades
| Grade | What it means | Example on this site |
|---|---|---|
| Grade A | A meta-analysis, or several randomized trials that agree, in people similar to the readers of the page. | Polarized and pyramidal training give the same race results (two meta-analyses). |
| Grade B | One randomized or controlled trial, a large cohort or big-data study with a consistent signal, or a formal expert consensus built on that kind of data. | A single run more than 10% longer than your recent longest run is linked to more overuse injuries (5,205 runners). |
| Grade C | Smaller observational studies, lab measurements that stand in for performance, or what elite athletes do in practice. | Norwegian double-threshold days (elite training logs, no trials). |
| Grade D | Coaching convention or expert opinion. The advice may be sensible, but its numbers have not been tested. | A long run of no more than about three hours. |
Some pages are marked Mixed when their parts rest on different levels of evidence, for example a ranking that combines trials with big-data studies.
The direction and the numbers can differ
A grade describes the main claim. The exact numbers in practical advice, such as how many weeks to taper or how much to cut, are often coaching convention even when the general direction is well supported. We say so on the page when that is the case, and we mark our own practical suggestions as suggestions.
How we choose sources
- We prefer recent meta-analyses and randomized trials, then large cohort studies, then smaller studies and expert consensus.
- We look for evidence in recreational runners, not only elite athletes or other sports, and we say when a result comes from somewhere else.
- Big-data studies of race results and sports-watch logs show links, not proof of cause. We say that each time.
- We link every study we cite, so you can read it yourself.
- Every page has a "What we don't know" section and a date for when it was last reviewed.
What we will not do
- Change a verdict because of an advertiser or an affiliate link.
- Present a popular book or method as proven when the evidence is thin, or dismiss it when the evidence supports it. Our verdicts are about specific claims, not about authors.
- Give personal medical advice. Chest pain, fainting, unusual breathlessness or confusion during or after running: stop and get medical help.
Where the research comes from
The first pages are based on a review of endurance-training research from 2020 to 2026, together with the older studies that recent work builds on. Much of that evidence is observational, comes mostly from men, and uses time trials shorter than a marathon. Where that limits a verdict, the page says so.