Autogenic training (AT) is a self-relaxation procedure by which a psychophysiological determined relaxation response is elicited. A meta-analysis was performed to evaluate the clinical effectiveness of AT. Seventy-three controlled outcome studies were found (published 1952-99). Sixty studies (35 randomized controlled trials [RCT]) qualified for inclusion in the meta-analysis. Medium-to-large effect sizes (ES) occurred for pre-post comparisons of disease-specific AT-effects, with the RCTs showing larger ES. When AT was compared to real control conditions, medium ES were found. Comparisons of AT versus other psychological treatment mostly resulted in no effects or small negative ES. This pattern of results was stable at follow-up. Unspecific AT-effects (i.e., effects on mood, cognitive performance, quality of life, and physiological variables) tended to be even larger than main effects. Separate meta-analyses for different disorders revealed a significant reduction of the heterogeneity of ES. Positive effects (medium range) of AT and of AT versus control in the meta-analysis of at least 3 studies were found for tension headache/migraine, mild-to-moderate essential hypertension, coronary heart disease, asthma bronchiale, somatoform pain disorder (unspecified type), Raynaud's disease, anxiety disorders, mild-to-moderate depression/dysthymia, and functional sleep disorders.