The effects of a social skill training program on socially isolated children are reported. The training program was derived from the results of a previous study of the specific social skills that discriminated popular from unpopular children. Training effects in the present investigation were assessed on sociometric position, and on the quality, frequency, and distribution of social interaction to peers. Time-series analyses were performed on the frequency and distribution of peer interaction. Two isolated children received social skill training with an adult coach, and two isolated children spent an equivalent amount of time with an adult. Results indicated that socially isolated children in the treatment group changed significantly in sociometric position on a follow-up assessment 9 weeks after the end of the intervention, did not change in the total frequency of peer social interaction, but did redistribute their interaction to peers. The two isolated children who did not receive the treatment program did not change significantly on the total frequency of peer interaction, and tended to withdraw from peers rather than redistribute their interaction. The insensitivity of total peer interaction frequency as a measure of outcome was discussed.