The purposes of this study were to (1) compare peer and self-video modeling in terms of effectiveness and efficiency in teaching first aid skills to children with intellectual disability and (2) analyze the error patterns made in probe sessions to determine whether the children who took the role of sufferers during the first aid skill sessions acquired the first aid skills by observing their peers and whether the research findings have social validity. Participants consisted of three children (two females and one male) between the ages of nine and 14 who have intellectual disability. An adapted alternating treatments design was used to compare the effects of peer and self-video modeling. According to the data, both peer and self-video modeling are equally effective and efficient. The errors made by the children mostly consisted of sequential and topographical errors. The children who played the role of sufferers acquired both skill sets by only observing their peers. The social validity data is positive. Results, implications, and opportunities for future research are discussed.