Objectives Comparing the overall and differential effects of psychodynamic psychotherapy (PDT) versus cognitive behavior therapy (CBT) for social anxiety disorder (SAD). Design Patients with a primary SAD (N = 47) were randomly assigned to PDT (N = 22) or CBT (N = 27). Both PDT and CBT consisted of up to 36 sessions (average PDT 31.4 and CBT 19.8 sessions). Assessments took place at waitlist: pretest, after 12 and 24 weeks for those who received longer treatment: posttest, 3‐month and 1‐year follow‐up. Methods Changes in the main outcome measure self‐reported social anxiety composite, as well as in other psychopathology, social skills, negative social beliefs, public self‐consciousness, defense mechanisms, personal goals, independent rater's judgments of SAD and general improvement, and approach behavior during an objective test, were analyzed using multilevel analysis. Results No improvement occurred during waitlist. Treatments were highly efficacious, with large within‐subject effect sizes for social anxiety, but no differences between PDT and CBT on general and treatment‐specific measures occurred. Remission rates were over 50% and similar for PDT and CBT. Personality disorders did not influence the effects of PDT or CBT. Conclusions PDT and CBT are both effective approaches for SAD. Further research is needed on the cost‐effectiveness of PDT versus CBT, on different lengths PDT, and on patient preferences and their relationship to outcome of PDT versus CBT.