We embed a model of the labor market with sector-specific search-and-matching frictions into a Ricardian model with a continuum of goods to show that trade liberalization causes higher unemployment in countries with comparative advantage in sectors with strong labor market frictions and leads to lower unemployment in countries with comparative advantage in sectors with weak labor market frictions. We test this prediction in a panel dataset of 97 countries during the period 1995-2009 and find that the data supports the theoretical prediction. Our results also help reconciliate the apparently contradicting evidence in the empirical literature on the impact of trade on unemployment.