We scrutinize the impact of international productivity gains (spillovers) induced by imports and exports on optimal tariffs. First, we solve a stylized 2x2 trade model of a large open economy and show that (a) productivity gains via exports and imports both reduce the strategically optimal tariff, (b) there exists a certain strength of productivity gains such that the incentive to manipulate the terms of trade strategically vanishes, (c) the welfare gain that can be achieved via a tariff is lower in the presence of productivity gains than in their absence, and (d) these results even hold without power on international markets. Second, we apply this model to a panel data set covering 40 countries, 29 sectors and the years 1995 to 2009. We find that importdriven productivity gains are stronger than export-driven productivity gains. Third, we extend our 2x2 model to a multi-region, multi-sector model that we calibrate to the data set used in the econometric analysis and to the econometrically estimated productivity gains. Optimal tariffs are reduced by 17% for the US and China and 40% for Brazil when taking trade-induced productivity gains into account. The USA are the only model region that gains from European optimal tariff policy. Thus, trade-induced productivity gains have empirically relevant effects on optimal tariffs.