Retraining is often hailed as a key policy tool to support workers displaced by import competition, yet there is surprisingly little evidence on whether these policies achieve their intended effects. Using administrative data from Germany, a highly open economy with extensive government-subsidized retraining programs, we provide evidence that workers routinely retrain in response to import competition. To quantify the welfare impact of retraining policies, we propose a search model in which heterogeneous workers may choose to retrain while unemployed. Retraining enables workers to change their job-finding rates and their productivity while employed. We find that retraining increases the gains from trade by 7% in the aggregate. Some worker groups gain five times as much, while others gain virtually nothing.