Objectives Mindfulness-based stress reduction (MBSR) is a structured 8-week group program teaching several types of mindfulness meditation techniques, including mindful yoga exercises. MBSR aims to help participants develop nonjudgmental awareness of moment-to-moment experience. Patients suffering from migraine usually benefit from behavioural and life-style interventions, but so far no mindfulness-based interventions were assessed. Methods We applied a new and innovative method termed design adaptive allocation to assign 62 patients to two study arms: (1) MBSR, (2) an active control procedure to account for nonspecific effects. The allocation procedures aimed at paralleling groups according to the predictors sex, age and migraine severity on the basis of a regression analysis and replaces classical randomization. The main outcome criterion of the trial is number of migraine days per month determined by a headache diary. Secondary outcome variables are quality of life, mental health, pain intensity, pain acceptance, medication and mindfulness. Results Thirty patients were allocated to the control group and 32 to the MBSR intervention. t -Tests indicated that there was a high degree of balance for sex (male 9.4% vs. 10%, p =0.935), age (44.8 vs. 44.5 years, p =0.918), and migraine severity (5.94 vs. 5.93 migraines, p =0.994). Conclusion It could be shown that this new allocation method guarantees extremely similar groups even with small samples. Our data prove that it is possible to avoid baseline differences in small pilot trials. Thus design adaptive allocation shows an advantage over the classical randomization procedures which often results in unbalanced groups when samples are small.