We develop a general equilibrium model of trade that features "indirectly additive" preferences and heterogeneous firms. Monopolistic competition generates markups that are increasing in firm productivity and in destination country per capita income, but independent from destination population, as documented empirically. The gains from trade liberalization are lower than in models based on CES preferences, and the difference is governed by the average pass-through. When we calibrate the model so as to match observed pricing-to-market in micro-data, it generates welfare gains that are substantially lower than those predicted by commonly employed frameworks.