To test the hypothesis that the social nature of the pupil team-learning situation differs in a number of respects from nongraded and conventionally graded classroom situations, sociometric response data were analyzed from 581 students, comprising 20 classrooms in grades 7 through 12 in four western New York rural schools. The sociometric device provided for subjective and perceptive selections on the dimensions of "liking,""school competence," and "social power." The study found that greater diffusion of sociometric choices and perception of choices made by others were associated with membership in nongraded and conventionally graded classrooms than in pupil team-learning graded and pupil team-learning nongraded classrooms. This result was attributed mainly to instruction occurring at the team level with the teacher providing instruction to individual pairs. Based on an analysis of data from the classes in social studies, English, and mathematics, it was concluded that subject studied as well as class size had no appreciable effect on the sociometric choices of students. (JK)