Despite the lengthy history of team cohesion-performance research, little is known about their reciprocal relationships over time. Using meta-analysis, we synthesize findings from 17 CLP design studies, and analyze their results using SEM. Results support that team cohesion and performance are related reciprocally with each other over time. We then used longitudinal data from 205 members of 57 student teams who competed in a complex business simulation over 10 weeks, to test: (a) whether team cohesion and performance were related reciprocally over multiple time periods, (b) the relative magnitude of those relationships, and (c) whether they were stable over time. We also considered the influence of team members' academic competence and degree of shared leadership on these dynamics. As anticipated, cohesion and performance were related positively, and reciprocally, over time. However, the cohesion → performance relationship was significantly higher than the performance → cohesion relationship. Moreover, the cohesion → performance relationship grew stronger over time whereas the performance → cohesion relationship remained fairly consistent over time. As expected, shared leadership related positively to team cohesion but not directly to their performance; whereas average team member academic competence related positively to team performance but was unrelated to team cohesion. Finally, we conducted and report a replication using a second sample of students competing in a business simulation. Our earlier substantive relationships were mostly replicated, and we illustrated the dynamic temporal properties of shared leadership. We discuss these findings in terms of theoretical importance, applied implications, and directions for future research.