Sherif (1958) hypothesized that competition between groups increases intergroup bias, cooperation decreases bias, and that cooperation's beneficial effects will be maximized when groups work together toward superordinate goals (i.e., goals which cannot be achieved without both group's contributions). The purpose of this dissertation is to test whether bias decreases when intergroup cooperation improves the probability of goal achievement (i.e., is instrumental) and increases bias when intergroup cooperation reduces the probability of goal achievement (i.e., is dysfunctional). Three studies investigated and supported this hypothesis. The first study used real world data from groups with a pre-existing history of conflict. The first study used Catholic and Protestant university students in Northern Ireland and assessed their perceptions of the instrumentality of intergroup relations and their feelings of warmth and forgiveness toward the religious outgroup. The second study established causality as well as distinguished instrumentality from the mutual intergroup differentiation model in a laboratory context. To establish causality the second study directly manipulated perceptions of instrumentality by explaining to participants that because of their respectively different thinking styles it would be instrumental or dysfunctional to work with another group. In a control condition no information was offered about instrumentality. To distinguish instrumentality from the mutual intergroup differentiation model, which emphasizes the importance that groups maintain distinctive identities during interactions, the second study also manipulated participants' abilities to differentiate the in- from the out-group. Although bias was lowest under instrumental conditions, bias actually increased when the groups were more clearly differentiated from each other during the cooperative interaction. The third study established factors that influence instrumentality and distinguished instrumentality from reinforcement. To assess factors that influence instrumentality, the third study manipulated the relative skill of the ingroup (high or low) and how the two groups' contributions would be combined (adding or averaging). When the intergroup performance was calculated by adding all participants' scores, working together was instrumental for each group. However, when the performance was calculated by averaging each person's contribution to derive the intergroup score, cooperation was dysfunctional (they would have preferred to work separately) for the highly skilled group as they could achieve a higher score working alone. The results revealed that intergroup bias was lower when cooperation was instrumental than when it was dysfunctional. To distinguish instrumentality from a reinforcement perspective, I assessed how much the instrumental conditions produced positive affect (which was assumed to be a proxy for reinforcement) and then assessed whether instrumentality had unique effects on bias controlling for reinforcement. Although there was no main effect of instrumentality on positive affect, instrumentality no longer related to intergroup bias when positive affect was statistically controlled. However, when controlling for positive affect there is a reliable negative relationship between perceived instrumentality and bias. Therefore, whether the effects of instrumentality on intergroup bias are completely independent of reinforcement is equivocal at this time.