I study individual location, education and work decisions in a dynamic life-cycle model in a developing country. I estimate the model exploiting panel data on migrants and stayers in Burkina Faso, and cross-sectional data on permanent emigrants. Individuals self-select into migration and locations based on education. Migration to urban centres increases with education, while migrants at the extremes of the education distribution tend to move abroad. Local unemployment rates, skilled work opportunities and returns to education result in differential expected income gains across locations and hereby explain the complex migration pattern observed. Large income gains from migration are partially offset by direct and indirect migration costs, as well as by higher investment in education (for rural migrants). Migration prospects to urban centres drive education choices of rural individuals. Hence, migration policies can be used to stimulate educational attainment in rural regions.