We assessed whether the stereotyped movements (SM) that are a defining characteristic of autism are discriminable from those observed in other disorders, and whether stereotyped self-injurious movements, which are excluded as exemplars of SM in DSM-IV, differ from other SM in severity or in kind. We used the Stereotyped and Self-Injurious Movement Interview to assess self-injurious and other SM in children with autism (n=56), intellectual disability (n=29), vision impairment (n=50), or hearing impairment (n=51) and in typical children (n=30). Cross-tabulation of scores indicated that self-injurious behavior is rarely performed in the absence of other SM. Reliability analyses indicated that patterns of covariation among SM items differ across groups so that different item sets are necessary to reliably measure SM in each group. Analyses of variance indicated the autism group exceeded one or more other groups in the frequency of 15 SM, the vision impaired group exceeded others on 5 SM, and the hearing impaired group exceeded others on 1 SM. Discriminant function analysis of SM items indicated that although only 66% of participants were accurately classified, it was rare for a child with a different disorder to be misclassified as having autism or visual impairment. We concluded that self-injurious behavior is a more severe form of SM, and there is a distinctive pattern of SM, including self-injurious behavior, that characterizes children with autism.