A study was conducted to test the hypothesis that adults' differential definitions of social competence for boys and for girls may contribute to lack of agreement between teachers' and classmates' ratings of children's social competence. Subjects were 540 children in fourth- and fifth-grade classrooms in four public schools in the greater Boston area. Two schools were upper middle class suburban, one was urban in a disadvantaged neighborhood, and one was suburban with a substantial population of recently arrived immigrants. Teachers were asked to nominate up to three children from their classrooms who matched a short paragraph describing three categories of social behavior: socially inhibited, aggressive/acting-out, and socially competent. Students in each class were administered the Minnesota Revision of the Class Play Inventory (CPI). The revised CPI yields a summary score on three behavioral categories: popular/leadership, aggressive/disruptive, and sensitive/isolated. The top three scorers in each behavioral category on the CPI and the three children nominated by their teachers were compared with respect to the sex of the children nominated. Teacher and peer nominations differed in that teachers, whether male or female, seldom assigned girls to the aggressive/acting-out category. Reasons for sex differences in teachers' ratings were explored and the importance of convergent measures of social competence was emphasized. (RH)