This study investigated the potential moderating effects of productivity norms and identifiability of effort on the cohesion-performance relationship in team sports. The design was a 2 (high cohesion, low cohesion) x 2 (high productivity norm, low productivity norm) x 2 (high identifiability of an individual's effort, low identifiability of an individual's effort) factorial. Each participant (n = 324) read one of eight scenarios, with cohesion, norms, and identifiability systematically rotated and indicated the probability that the individual would train during the off-season. An ANOVA showed a main effect for cohesion, F(1, 316) = 113.44 p < .0001, and norms, F(1, 316) = 19.61, (p < .0001), and an interaction between cohesion and norms, F(1, 316) = 735, p = .007 The probability of off-season training was significantly higher for the high-cohesion-high-norms scenario than for the high-cohesion-low-norms scenario, with no differences under conditions of low cohesion. Directions for future research are discussed.