Previous research has only recently begun to examine the effects of knowledge management system (KMS) usage on work group outcomes. As a result, little is known about the conditions under which the usage of repository KMS increases group performance. In response to this gap, this study examines the contingencies of group composition. We conceptualize repository KMS usage as a learning mechanism that positively impacts group performance. Further, we identify group size, knowledge heterogeneity, and membership change as aspects of group composition which moderate this relationship. We analyzed a unique dataset composed of objective measures of repository KMS usage and sales commissions of 3,092 work groups within a large franchise company. The results indicate that repository KMS usage has an overall positive effect on group performance. This impact is diminished for groups with a greater number of members, with greater knowledge heterogeneity, and with greater membership change. Our results suggest that the implementation of technology alone is not sufficient to generate value for groups. Instead, close attention should also be paid to barriers which inhibit group learning.