In order to generate information about the relative effectiveness of the most widely used pharmacological and non-pharmacological interventions for the prophylaxis of recurrent migraine (i.e., propranolol HCl and combined relaxation/thermal biofeedback training), meta-analysis was used to integrate results from 25 clinical trials evaluating the effectiveness of propranolol and 35 clinical trials evaluating the effectiveness of relaxation/biofeedback training (2445 patients, collectively). Meta-analysis revealed substantial, but very similar improvements have been obtained with propranolol and with relaxation/biofeedback training. When daily recordings have been used to assess treatment outcome, both propranolol and relaxation/biofeedback have yielded a 43% reduction in migraine headache activity in the average patient. When improvements have been assessed using other outcome measures (e.g., physician/therapist ratings), improvements observed with each treatment have been about 20% greater. In both cases, improvements observed with propranolol and relaxation/biofeedback have been significantly larger than improvement observed with placebo medication (14% reduction) or in untreated patients (no reduction). Meta-analysis thus revealed substantial empirical support for the effectiveness of both propranolol and relaxation/biofeedback training, but revealed no support for the contention that the two treatments differ in effectiveness. These results suggest that greater attention should be paid to determining the relative costs and benefits of widely used pharmacological and non-pharmacological treatments.