Twenty-two "socially dysfunctional" outpatients had 12 1 1/2-hour sessions of social skills training either alone or combined with cognitive modification (each, n = 11). Two raters who were blind to the treatment regimen significantly differentiated the two treatments, as judged from audiotapes of treatment sessions. Measurement was by an independent assessor in a structured interview, by self-monitoring of several activities and of anxiety felt during them, and by self-rated questionnaires. In each treatment regimen, patients' conditions improved significantly and equally on reported behavioral and cognitive measures both during and after treatment, and at six-month follow-up. Outcome did not differ significantly between the two treatment conditions at any time on any measure. Patients reported increased levels of social activities and less anxiety during these, less isolation, better relations with their colleagues at work, less depression, and loss of many irrational social beliefs. During multiple baseline monitoring for up to nine weeks, no improvement occurred. Performance of each social task or "target" increased significantly after training was begun for it, and subjective anxiety decreased similarly during social performance. Untreated targets improved much less, and more slowly. Social skills training was followed by lasting and worthwhile improvement that was not enhanced by the addition of cognitive modification.