In this article, we meta-analytically examined the efficacy of biofeedback (BFB) in treating migraine. A computerized literature search of the databases Medline, PsycInfo, Psyndex and the Cochrane library, enhanced by a hand search, identified 86 outcome studies. A total of 55 studies, including randomized controlled trials as well as pre-post trials, met our inclusion criteria and were integrated. A medium effect size (d =0.58, 95% CI=0.52, 0.64) resulted for all BFB interventions and proved stable over an average follow-up phase of 17 months. Also, BFB was more effective than control conditions. Frequency of migraine attacks and perceived self-efficacy demonstrated the strongest improvements. Blood-volume-pulse feedback yielded higher effect sizes than peripheral skin temperature feedback and electromyography feedback. Moderator analyses revealed BFB in combination with home training to be more effective than therapies without home training. The influence of the meta-analytical methods on the effect sizes was systematically explored and the results proved to be robust across different methods of effect size calculation. Furthermore, there was no substantial relation between the validity of the integrated studies and the direct treatment effects. Finally, an intention-to-treat analysis showed that the treatment effects remained stable, even when drop-outs were considered as nonresponders.