In a randomized trial for children with social phobia (SP), Social Effectiveness Therapy for Children (SET-C; a treatment consisting of exposure and social skills training) and fluoxetine were more effective than pill placebo in reducing social distress and behavioral avoidance, but only SET-C demonstrated significantly improved overall social skill and social competence. In the current study, the authors examined the specific social skills enhanced by SET-C using a recently developed coding schema. At posttreatment, children treated with SET-C displayed a more effective ability to manage the conversational topic (pragmatic social behaviors) and more appropriate motor movement, facial orientation, and posture (paralinguistic social behaviors) than children treated with fluoxetine or placebo. In contrast, children treated with fluoxetine displayed no more pragmatic or paralinguistic skill than children given a pill placebo. There were no group differences on ratings of voice volume and vocal inflection (speech and prosodic social behaviors). Furthermore, only children treated with SET-C improved from pre- to posttreatment on all three skill variables. Findings suggest that pharmacological interventions that only target reduction in anxious arousal may not have an impact on social skill deficits and may not be adequate to optimally treat SP. The relationship of social skill to social avoidance and the importance of social skills training to enhance social competence in the treatment of childhood SP are discussed.