This research evaluated a videotape modeling intervention for six preschool children who showed low levels of social involvement with peers in natural free-play settings. Observations across 14 weeks gave frequency and topographic data on social interactions and assessment of social involvement in play using the Parten Scale. Simultaneous observation of three children gave data on two comparison children for each target child. A multiple-baseline across-subjects design was used with two data analysis strategies. Averaged group data showed statistically significant increases in both interactions and social involvement in play. Visual analysis of the single-subject data, however, indicated no clear treatment outcome for two children, while four subjects showed a variable increase in social responding after viewing the modeling videotape. Session-by-session variability was a feature of the interaction rate measure for comparison children and for posttreatment phase data for target subjects.