An 8-week trial of 2 mg/kg/day of naltrexone (Trexan, now called ReVia) was conducted with four males with autism and two males with pervasive developmental disorders. Two dependent measures were employed: (a) direct behavioral measurement conducted by blind observers using videotape recordings of periodic learning task sessions, and (b) performance measures of the learning activity. Direct behavior measures revealed one participant's attending-to-task improved while receiving naltrexone. However, two of the participants' rates of self-injury increased, as did one of those participant's rate of aggression. Significant improvement was observed on performance measures of the learning task; however, the improved performance continued during the non-drug return to baseline for the three participants who experienced this additional condition. Discussion is provided relating these findings to previously published results.