A number of recent analyses of the nuclear arms race have emphasized the role that psychological processes play in nations’ decisions to arm themselves. The present research was designed to investigate several hypotheses regarding how to cognitive processes, decision framing and the level of social differentiation in a group of interdependent decision makers, affects their decision to allocate economic resources to increase security. It was predicted that decision makers would allocate more resources to security (arm themselves at higher levels) when decision makers would allocate more resources to security (arm themselves at higher levels) when decision frame implied high deficits in security and when the perceived level of social differentiation in their groups was high. To test these hypotheses, two experiements were conducted using a laboratory simulation of an arms race. The results of both experiments provided strong support for the decision framing hypothesis. Support for the social differentiation hypothesis was more equivocal. Social differentiation based upon perceptions of “common fate: had inconsistent effects on behavior, while differentiation based upon naturally-ocurring social categories affected decision making as predicted. The external validity of the findings is discussed.