Hand and computer searches located studies on the effects of relaxation techniques on trait anxiety. Effect sizes for the different treatments (e.g., Progressive Relaxation, EMG Biofeedback, various forms of meditation, etc.) were calculated. Most of the treatments produced similar effect sizes except that Transcendental Meditation had significantly larger effect size (p less than .005), and meditation that involved concentration had significantly smaller effect. Correlations with effect size were calculated for many variables, e.g., population, age, sex, experimental design, duration and hours of treatment, pretest anxiety, demand characteristics, experimenter attitude, type of publication, attrition, etc. Only a few variables (mainly population, duration, hours, and attrition) significantly influenced effect size. Controlling for possible confounding variables did not alter the overall conclusions. The difference in effect size between treatments was maintained both when only published studies were included and when only the studies with the strongest design were included. Possible explanations for the findings are examined.