Team cohesion is an important antecedent of team performance, but our understanding of this relationship is mired by inconsistencies in how cohesion has been conceptualized and measured. The nature of teams is also changing, and the effect of this change is unclear. By meta-analyzing the cohesion-performance relationship (k = 195, n = 12,023), examining measurement moderators, and distinguishing modern and traditional team characteristics, we uncovered various insights. First, the cohesion-performance relationship varies based on degree of proximity. More proximal measures –task cohesion, referent-shift, and behaviorally-focused– show stronger relationships compared to social cohesion, direct consensus, and attitudinally-focused, which are more distal. Differences are more pronounced when performance metrics are also distal. Second, group pride is more predictive than expected. Third, the cohesion-performance relationship and predictive capacity of different measures are changing in modern contexts, but findings pertaining to optimal measurement approaches largely generalized. Lastly, important nuances across modern characteristics warrant attention in research and practice.