Concepts from bargaining theory and neorealist theory are integrated in a partners-and-rivals (PAR) model that explains the terms of collaboration between rational actors (for example, firms) that are potential competitors. The model contends that the stronger player's incentive to collaborate is determined not only by its interest in absolute welfare benefits from collaboration but also by its concern about and loss of utility from unfavorable shifts in relative position vis-à-vis the weaker player. Two propositions are derived from the model. According to the disparity principle, the stronger player's net payoff from collaboration (the sum of its welfare benefits and positional costs) is a function of the disparity in capabilities between the two players. The net payoff curve is low when the disparity in capabilities is large, reaches an optimum when the disparity is moderate, and then falls again as the disparity approaches zero. The slope of the curve is also affected by a coefficient, α, which reflects the stronger player's sensitivity to positional losses. According to the compensation principle, for collaboration to arise, the weaker player must make an adjustment in the terms of the bargain, or a side-payment, to compensate the stronger player for its positional losses from collaboration. The validity of the PAR model is tested by comparing Franco-German collaboration on military aircraft in two cases, one in which the disparity in capabilities between the players was moderate and the other in which the disparity was small.