A 12-year-old girl, hospitalized on a psychiatric intensive care unit, was treated for phobias of school and unfamiliar males. Treatment focused on decreasing several avoidance responses and increasing prosocial approach responses. Treatment consisted of instructions, performance feedback, participant modeling and social reinforcement. Marked changes were evident when treatment was introduced in a multiple-baseline design across several phobic and prosocial behaviors. Treatment effects generalized beyond the persons and situations included in training, were reflected in overall global ratings by persons unfamiliar with the treatment, and were maintained up to a 7-week follow-up when the child returned to the hospital for reassessment. Contact with the child 21 weeks after discharge indicated that the gains were reflected in school attendance and social interaction in everyday situations.