In a political environment increasingly concerned with the problem of escalating health care costs, the issue of psychotherapy efficacy evaluation is still debated. Historically, this debate started after Eysenck (1952) published a famous article suggesting that average improvement from pre- to post-therapy has nothing to do with psychotherapy participation, but something that would tend to happen anyway ("spontaneous remission"). Wanting to prove that psychotherapy was effective (Glass, 2000), Smith and Glass (1977) published the first meta-analysis of the psychotherapy efficacy, combining the results of several controlled clinical trials, and found that psychotherapy was remarkably efficacious. Following a brief historical introduction, the objectives of this paper is to define the principles of meta-analysis, to discuss of epistemological contextualization of this methodological approach, and finally to examine the interest and limits of the application of this method in the field of evaluation of psychotherapy.