Children's success or failure at interacting with peers has been conceptualized as a consequence of their particular competencies in social information processing. Dodge and colleagues have proposed a useful model with five sequential and separable steps, ranging from (1) encoding of social cues, to (5) response enactment (Dodge, Pettit, McClasky, & Brown, 1986). They used a novel videotape technique, requiring children to watch child actors interact in a scripted "peer group entry" situation; the videotape was stopped at various points in the interaction, and children's social information processing was probed at various points in the proposed cognitive sequence. While appropriate for assessing social cognition (Steps 2 through 5), this method does not seem as well suited for assessing individual differences in social perception (Step 1) because the videotapes do not contain spontaneous patterns of body kinematics and facial gesture which are representative of the structure of natural interactions. Videotapes of naturally occurring social interactions have been used in two recent studies of social perception (Van Acker & Valenti, 1989; Valenti & Wagner, 1991), showing reliable and above-chance accuracy of intention identification, and evidence that both facial and body gesture conveys information about children's intentions.