In organizational theory, it is a widely accepted postulate that cooperation among subjects is enforceable. This assumption is essential for the evaluation of two frequently discussed incentive systems: team and tournament compensation. Whereas in team-based pay systems cooperation is highly desired, cooperation in rank-order tournaments—labeled as ‘collusion’—is regarded as one of the main drawbacks of relative performance evaluation. In this experimental study, two different communication technologies are introduced into both incentive environments. The results indicate that when only limited communication is permitted subjects tend to cheat on each other in the tournament rather than in the team setting. Interestingly, allowing subjects to exchange emails leads to a similarly stable cooperation in both incentive systems. Copyright © 2006 John Wiley & Sons, Ltd.