International agreements are an important element of the world economic system, but questions remain as to why they are necessary. In trade theory, the terms-of-trade hypothesis posits that countries use tariffs in part to improve their terms of trade and that trade agreements cause them to internalize the costs that such terms-of-trade shifts impose on other countries. As yet, however, the hypothesis is largely untested on the most-favored-nation (MFN) tariff schedules of the vast majority of WTO members, which cover the bulk of world trade. This paper investigates whether MFN tariffs set by existing WTO members in the Uruguay round are consistent with the terms-of-trade hypothesis. We present a model of multilateral trade negotiations featuring endogenous participation that leads the resulting tariff schedules to display terms-of-trade effects. Specifically, the model predicts that the level of the importer's tariff resulting from negotiations should be negatively related to the product of exporter concentration, as measured by the Herfindahl-Hirschman index (sum of squared export shares), and the importer's market power, as measured by the inverse elasticity of export supply, on a product-by-product basis. We test this hypothesis using data on tariffs, trade and production across more than 30 WTO countries and find strong support. We estimate that the internalization of terms of trade effects through WTO negotiations has lowered the average tariff of these countries by 22 to 27 percent compared to its non-cooperative level.