Background: Migraine constitutes a good model for the study of placebo response. It is a well-defined disease, affects a large population and a great number of clinical trials have been performed, which have given homogeneous outcomes. Aim: The aim of this meta-analysis is to evaluate the placebo response rate in migraine prophylaxis in all published clinical trials since 1988 and to estimate the influence of study design in response variability. Methods: A computer-based information search was conducted on the Medline database. The outcomes studied were patients who improved (reduction in migraine attacks of 50% or more); attacks per month, and patients with adverse events. Study design and countries in which the study was carried out were also analysed. The meta-analysis was computed using the Mantel-Haenszel test. Results: In the final analysis, 32 papers were considered. The pooled estimate of the placebo response (patients who improved) was 21%. The placebo response rates were significantly higher in studies with a parallel design than those in cross-over studies (p < 0.01). This response was also higher in European studies than in those performed in North America (p < 0.001). Adverse events occurred in 30% of the patients who took a placebo, and the percentage of patients with adverse events was significantly higher in the North American studies than in those conducted in Europe (p < 0.01). Conclusion: These data reinforce the need to consider the placebo effect when ascertaining the true therapeutic effect of any drug, as well as to design any clinical trial in the prophylaxis of migraine. © 2007 European Federation of Chapters of the International Association for the Study of Pain.