Specifically designed to avoid methodological problems limiting earlier research, this study attempted to identify specific behavioral differences between elementary school students with either high or low sociometric status. Subjects, who had been selected on the basis of same-sex sociometric status nominations of 144 classmates, were 30 females and 24 males in seven fourth- and fifth-grade classrooms. The 54 subjects were observed on a quasi-random basis by 10 experimentally naive undergraduate students using a 31-category comprehensive state/event observation code. Each subject was observed in both academic and nonacademic settings for a minimum of 12 10-minute sessions. Results indicated an interaction between sociometric status and setting, but students of high and low sociometric status could be discriminated on the basis of behavior in both settings. It was tentatively concluded that the utilization of the overall rate of peer social interaction as an index of social skill or appropriate social behavior is not supported by the available data. Additionally, results suggested that certain behaviors frequently targeted as social skills are appropriate behavioral goals only for certain situations and that, to be understood, behavioral differences should be considered within the context of the behavioral setting in which they occur. (Implications for training in social skills are briefly considered.) (RH)