The gravity model of trade states that the volume of trade between two countries is proportional to the product of the sizes of the two countries and the inverse of the distance between them. The gravity model, however, was initially suggested for other types of social interactions, and it also predicts well the probability of militarized disputes. This raises a concern about omitted-variable bias in the literature on trade and conflict: if the gravity model is correlated with both trade and conflict, it is necessary to control for the size of both countries in the dyad. In this article, I analyze the nature of this bias both empirically and theoretically. I find that a correct model specification leads to a stronger relationship between trade and conflict than in the traditional models. However, the empirical finding must be reinterpreted: what is crucial is not the relationship between the value of the bilateral trade and the size of the larger economy in the dyad (‘trade dependence’), but rather ‘trade efficiency’—the extent to which the two countries trade more than expected from the gravity model.