The Behavior Summarized Evaluation (BSE), developed for the assessment of autistic behavior, was specifically designed to evaluate the severity of behavioral problems in autistic children involved in bioclinical and therapeutic studies. The reliability studies and the factorial analysis of this scale have been previously published. The present paper examines the effectiveness of the BSE to discriminate 58 autistic from 58 nonautistic mentally retarded children. The BSE clearly separated the two samples of children. A most efficient combination of 8 items emerged from the stepwise item selection procedure. The between-group differences were highest on 4 items, indicating that the most particular pattern in autistic compared to nonautistic children could be the association of autistic withdrawal and stereotypic behaviors. Our findings suggest that the BSE could help in the detection and evaluation of autistic developmental deviance. Implications for further research are discussed.