The participation in certain classroom processes by students of high and low mathematics confidence who scored above the mean in mathematics achievement was studied over 2 years. The processes selected were: (1) specified types of teacher-pupil interactions, and (2) student engaged time in high or low cognitive level mathematical activities, spatial activities, and with peers. Eighty-two seventh graders were observed daily in their regular mathematics classes for 3 to 4 weeks during the spring semester of 1980, and were again observed in 1981. Between 3 and 14 target students were within each observed class. Roughly equal numbers of each sex were chosen based on the following characteristics for the sample: (1) mathematics achievement scores were at or above the sixth-grade mean of four middle schools, and (2) confidence in mathematics scores were in either the top or bottom quarter for all pupils who had achieved higher than the mean. Two trained observers recorded data on target student and teacher behavior in each classroom. Data were collapsed across classrooms and analyzed using analysis of variance, with sex, confidence level, and year as factors. (MP)