<p>We develop a theoretical framework and conduct a laboratory experiment to study how group identity affects cooperative behavior and strategy selection in infinitely repeated prisoner’s dilemma (IRPD) games. We find that participants are more likely to cooperate and less likely to adopt the Always Defect strategy with ingroup members than with outgroup members or participants in the control condition. Ingroup pairs are also more likely to sustain cooperation but less likely to persist in defection over the dynamic course across supergames, compared to both outgroup and control pairs. However, these effects are statistically significant only in the low strategic risk environment when the discount factor is high enough for cooperation to be both a subgame perfect Nash equilibrium and risk dominant. In the high strategic risk environment where cooperation is a subgame perfect Nash equilibrium but not risk dominant, the impact of group identity on cooperation is less robust and only holds qualitatively.</p>