This talk challenges apparently simple notions of when to do (or not to do) randomised controlled clinical trials.