Although role or perspective taking has been considered important for the development of mature social thought and behavior, and training for perspective taking has been used to remediate deficient behavior, few studies with either normal or disturbed populations have examined naturally occurring behaviors expected to correlate with perspective-taking ability. This study examines the relationship between perspective taking and the social behavior and affective language of 26 behaviorally disordered boys ages 5 to 12. Each subject was observed for approximately 33 minutes in lunch, freetime, and organized activity settings varying in level of structured expectations for behavior. The 12 behavior categories used in coding were designed to record the frequency and quality of the child's interaction with others, both adults and peers. Measures of perspective taking, affective language, and receptive vocabulary were administered. Instruments employed were Chandler's (1973) role-taking task; the interview from the friendship domain of the Measure of Interpersonal Understanding developed by Selman, Jaquette, and Bruss-Saunders (1979); and the Peabody Picture Vocabulary Test. Results indicated that perspective-takers spent less time "alone" on-task, received less attention from adults, and shared more than their non-perspective-taking peers. Perspective-takers also spent more time in neutral interaction with other children than did non-perspective-takers. Finally, perspective-takers used a greater variety of affective words in response to the Chandler role-taking measure. (Author/RH)