Deficit in social interaction is a primary component of infantile autism. However, in the majority of drug studies, social interaction has not been measured consistently over time. Therefore, we examined, in a crossover design, the effect of fenfluramine on the social interactions of seven autistic children. Social interaction was measured one to three times per week, while the children were in open placebo, placebo, or drug phases of the study. The results demonstrated that the effect of fenfluramine on social interaction was inconsistent across children, with two children possibly demonstrating a tolerance to the behavioral effects of the drug. The results are discussed with respect to genetic and pharmacologic factors.