A 2 (committed or uncommitted) × 2 (high power or low power) × 3 (structure: Prisoner's Dilemma or Chicken or No Conflict) factorial design was used to examine the effects of commitment to future interaction, relative power, and the structure of interdependence. The dependent variable was the proportion of cooperative choices on a series of single-play game matrices. Data were collected from 120 undergraduate students in a laboratory situation. The task consisted of responding either cooperatively or competitively to 18 game matrices. Each matrix was presented individually and without feedback concerning previous outcomes. As hypothesized, the results showed an interaction between commitment and power. Low reward power subjects were more cooperative when uncommitted than when committed in Chicken and the No Conflict game. High power subjects were more cooperative when committed than when uncommitted in the Prisoner's Dilemma. These results suggest caution when generalizing from the prevalent propositions that low power and commitment to future interaction have a cooperative effect.