The purpose of this study was to investigate the effects of a modeling and self-monitoring intervention on the play skills and social initiations of two children with autism spectrum disorders (ASD). This study was conducted in a therapy room of a hospital located in Seoul city. A multiple baseline design across settings that included watercolors, Play-cone, and Playdough was employed. The experimental procedures were as follows: Child 1 had baseline, modeling and self-monitoring intervention, self-monitoring intervention only, and maintenance; Child 2 had baseline, modeling and self-monitoring intervention, modeling & self-monitoring with edible reinforcement, self-monitoring intervention only, and maintenance. During baseline, the children were asked to play without any intervention and observed for 10 min. If the children did not play for one min or showed problem behaviors related to the safety, prompts were provided. In the modeling and self-monitoring intervention phase, the researcher provided the children with modeling to teach play skills and play-related social initiation behaviors. If the children did not respond within 5s after the modeling, systematic prompting was provided until a correct response occurred. At the completion of each session a reinforcer was provided based on the self-monitoring system. All procedures were employed in the same way to two children except that an edible reinforcer was added to the Child 2. In the self-monitoring intervention, modeling was removed and only self-monitoring was used. Finally, in maintenance, all interventions were removed and the children''s play skills and social initiation behaviors were monitored and measured for 10 min in the same manner as in baseline. In each experimental procedure, 25% of sessions were randomly selected to measure inter-observer agreement and procedural fidelity. In addition, social validity of the intervention was assessed by the mothers of two participants. The results of the study indicated that the modeling and self-monitoring intervention was effective in increasing spontaneous play skills and play-related social initiations of both participants. In addition, the participants maintained the increased play skills after removing the intervention. In terms of the social initiations, Child 1 maintained increased social initiations after the removal of the intervention; however, the social initiations of the Child 2 decreased. The mothers of the children who participated in this study reported that the modeling and self-monitoring intervention used in the study was effective in improving spontaneous play skills.