Recently, game theoretical models have become a popular tool among students of human and non-human social behavior. Social interactions among "players" in a game can often be divided into three phases; assortative, interactive and allocative. During the assortative phase, individuals decide who to interact with, during the interactive phase, how to interact and during the allocative phase, how to divide the spoils. In most of these models, strategic behavior is employed only during the interactive phase, while pair formation during the assortative phase is assumed to be random. In many social interactions, however, pair formation is not random during the assortative phase and individuals often exhibit active "partner choice". Here, I will examine the evidence, both theoretical and empirical, that partner choice may play an important role in social interactions. Evidence for non-random pair formation is presented in the context of mate choice, cooperative behavior, aggressive interactions, predator recognition and foraging. Active partner choice often facilitates the evolution of cooperative and altruistic behavior and appears to be common in non-human animals.