This paper uses confidential U.S. Customs data on U.S. importers and their Chinese exporters toinvestigate the frictions from changing exporting partners. High costs from switching partners can affect the efficiency of buyer-supplier matches by impeding the movement of importers from high to lower cost exporters. I test the significance of this channel using U.S. import data, which identifies firms on both sides (U.S. and foreign) of an international trade relationship, the location of the foreign supplier, and values and quantities for the universe of U.S. import transactions. Using transactions with China from 2003-2008, I find evidence suggesting that barriers to switching exporters are considerable: 45% of arm’s-length importers maintain their partner from one year to the next, and one-third of all switching importers remain in the same city as their original partner. In addition, importers paying the highest prices are the most likely to change their exporting partner. Guided by these empirical regularities, I propose and structurally estimate a dynamic discrete choice model of exporter choice, embedded in a heterogeneous firm model of international trade. In the model, importing firms choose a future partner using information for each choice, but are subject to partner and location-specific costs if they decide to switch their current partner. Structural estimates of switching costs are large, and heterogeneous across industries. For the random sample of 50 industries I use, halving switching costs shrinks the fraction of importers remaining with their partner from 57% to 18%, and this improvement in match efficiency leads to a 12.5% decrease in the U.S.-China Import Price Index.