This paper examines income maximization and the selection and sorting of international migrants. Using data on emigrant stocks by schooling level and source country in OECD destinations, the authors find that a simple model of income maximization can explain both positive selection (more educated migrants are more likely to emigrate) and positive sorting (more-educated migrants are more likely to settle in destination countries with high rewards to skill). Results show that migrants are more educated relative to non-migrants when the absolute skill-related difference in earnings between the destination and source is larger. The relative stock of more-educated migrants in a destination increases with the absolute earnings difference between high and low-skilled workers. The authors also estimate the magnitude of migration costs by source-destination pair and assess the contribution of wage differences to how migrants sort themselves across destination countries. The paper develops and estimates a simple model of migration based on the Roy (1951) income maximization framework. The model predicts that an increase in the reward to skill in a destination should cause immigration from source countries to rise and the mix of migrants to become more skilled. The model delivers estimating equations for the scale of migration, the selection of migrants in terms of schooling, and the sorting of migrants across destinations by schooling. The three equations estimate a common coefficient on earnings, but differ in terms of the data they require and the assumptions one must impose regarding migration costs. The authors use data from Beine, Docquier, and Rapoport (2007) on the stock of migrants by education level from 192 source countries residing in OECD destination countries as of 2000. The data strongly support income maximization. In the scale regression, migration is increasing in the level earnings difference between the destination and the source, although the estimated effect of earnings appears to be attenuated due to omitted fixed costs of migration. In the selection and sorting regressions, which difference out fixed costs, the relative stock of more-educated migrants is larger in destinations with greater skill-related earnings differences. The authors also find post-tax earnings are a stronger correlate of migration than pre-tax earnings, consistent with migrants weighing tax treatment. The paper addresses conflicting results on migrant selectivity. Borjas (1987) develops a version of the Roy model which predicts that migrants who move from a country with high returns to skill to a destination with low returns to skill should be negatively selected. However, the Borjas (1987) framework performs well in explaining migration from Puerto Rico to the U.S. (Ramos, 1992; Borjas, 2006), but less well elsewhere. Migrants from Mexico to the U.S. are drawn from the middle of the skill distribution, even though returns to skill are higher in Mexico than the US. The authors find that OECD-bound migrants are positively selected, even though many are from countries where returns to skill exceed those in the OECD. The authors also estimate an alternativeThis paper examines income maximization and the selection and sorting of international migrants. Using data on emigrant stocks by schooling level and source country in OECD destinations, the authors find that a simple model of income maximization can explain both positive selection (more educated migrants are more likely to emigrate) and positive sorting (more-educated migrants are more likely to settle in destination countries with high rewards to skill). Results show that migrants are more educated relative to non-migrants when the absolute skill-related difference in earnings between the destination and source is larger. The relative stock of more-educated migrants in a destination increases with the absolute earnings difference between high and low-skilled workers. The authors also estimate the magnitude of migration costs by source-destination pair and assess the contribution of wage differences to how migrants sort themselves across destination countries. The paper develops and estimates a simple model of migration based on the Roy (1951) income maximization framework. The model predicts that an increase in the reward to skill in a destination should cause immigration from source countries to rise and the mix of migrants to become more skilled. The model delivers estimating equations for the scale of migration, the selection of migrants in terms of schooling, and the sorting of migrants across destinations by schooling. The three equations estimate a common coefficient on earnings, but differ in terms of the data they require and the assumptions one must impose regarding migration costs. The authors use data from Beine, Docquier, and Rapoport (2007) on the stock of migrants by education level from 192 source countries residing in OECD destination countries as of 2000. The data strongly support income maximization. In the scale regression, migration is increasing in the level earnings difference between the destination and the source, although the estimated effect of earnings appears to be attenuated due to omitted fixed costs of migration. In the selection and sorting regressions, which difference out fixed costs, the relative stock of more-educated migrants is larger in destinations with greater skill-related earnings differences. The authors also find post-tax earnings are a stronger correlate of migration than pre-tax earnings, consistent with migrants weighing tax treatment. The paper addresses conflicting results on migrant selectivity. Borjas (1987) develops a version of the Roy model which predicts that migrants who move from a country with high returns to skill to a destination with low returns to skill should be negatively selected. However, the Borjas (1987) framework performs well in explaining migration from Puerto Rico to the U.S. (Ramos, 1992; Borjas, 2006), but less well elsewhere. Migrants from Mexico to the U.S. are drawn from the middle of the skill distribution, even though returns to skill are higher in Mexico than the US. The authors find that OECD-bound migrants are positively selected, even though many are from countries where returns to skill exceed those in the OECD. The authors also estimate an alternative