This study investigated the effects of group ability composition (homogeneous versus heterogeneous) on group processes and outcomes for high ability students completing science performance assessments. Participants were 99 seventh and eighth graders from 9 classes in 2 schools. The results show that group ability composition does not have straightforward effects on achievement. While high-ability students working in homogeneous groups uniformly performed well, high-ability students in some heterogeneous groups performed better than high-ability students in other heterogeneous groups. The quality of group functioning served as the strongest predictor of high-ability students' achievement. High-ability students in groups that were responsive to group members' needs for help and did not engage in debilitating social behavior performed well, whereas high-ability students in groups with poorer functioning did not. While homogeneous groups consistently showed beneficial group functioning, only some heterogeneous groups exhibited these traits. These results show that achievement of high-ability students cannot be predicted from a simple homogeneous-heterogeneous grouping contrast and that the level of group functioning provides the key to understanding group performance. An appendix contains sample test questions. (Contains 10 tables, 10 figures, and 63 references.) (SLD)