Psychological group treatments, such as behavioral or cognitive-behavioral therapy, are generally effective interventions for social phobia. However, a substantial number of individuals discontinue these treatments prematurely. Participant attrition can threaten the validity of treatment outcome studies if attrition during therapy does not occur randomly. In order to examine this issue, we studied 133 individuals with a principal diagnosis of social phobia who initiated a 12-week behavioral or cognitive-behavioral group treatment for social phobia. Thirty-four participants discontinued therapy prematurely. These dropouts were compared to treatment completers in demographic characteristics, Axis I and II psychopathology, and their attitude toward treatment. The results only showed a small difference between treatment completers and dropouts in their attitude toward treatment: dropouts rated the treatment rationale as less logical than completers at the beginning of treatment. No other differences between dropouts and completers were observed. Therefore, dropouts are unlikely to present a serious threat to the external validity of treatment outcome studies for social phobia.