There is a long-standing polemic concerning the usefulness of antidepressants in the treatment of major depressive disorder. In this paper, we propose to highlight some aspects of this controversy by exploring the mutual influence of psychopharmacology and trial methodologies. Indeed, antidepressant efficacy, if not proved, was accepted before antidepressant randomised controlled trials (RCTs) were run. While RCTs became a gold standard to meet the requirements of the regulatory bodies, methodological tools were required to measure outcomes and to test whether antidepressants provide statistically significant benefits as compared with a placebo. All these methodological options have nonetheless introduced fuzziness in our interpretation of study results, in terms of clinical meaningfulness and in terms of transposability to a real life settings. Additionally, selective publication raises concerns about the published literature, and results in many paradoxes. Instead of providing easy answers, the application of the RCT paradigm in MDD raises numerous questions. This is probably in the nature of all scientific studies, but it can be in contradiction with clinicians' expectations, who want to be sure that the treatment will (or will not) work for their individual patients.