This attempt to establish treatment effect sizes follows a tradition, which dates back over 20 years, of assessing the effects of antidepressant treatment under ‘blinder conditions'. The method adopted by Moncrieff et al has some merit but involves a recourse to studies, many of which are over 30 years old. While the studies may not be seriously flawed, it is difficult to have much confidence in them. None appears to have included what the authors describe as an ‘inert’ placebo, which strictly speaking would be a non-drug arm to the study. The number of studies is small. The dose of drugs used, which might be expected to have some influence on outcomes, is not mentioned. Finally, as experience with studies in obsessive-compulsive disorder indicates, treatment effect sizes can vary substantially from one decade to another –most probably because different individuals are recruited although all may apparently meet the same diagnostic criteria.