Abstract A model is constructed in which multinational firms arise endogenously. Multinationals are more important in total activity when countries are similar in incomes (size) and in relative factor endowments, and when total world income is high. These predictions are consistent with empirical evidence, and our results help point to more formal tests. The standard oligopoly model of international trade is a special case of our model when multinationals are suppressed, and this allows us to provide an explicit comparison to the national-firm model with respect to the location of production, welfare, and the volume of trade.