Casual evidence for some developed countries suggests that most talented migrants become entrepreneurs (positive sorting), but entrepreneurship might also be chosen by less talented migrants who have fewer opportunities in the labour market of the destination countries (negative sorting). Building upon Lucas (1978), we develop a theoretical framework to analyze the different mechanisms at play that draw migrants into entrepreneurship. The model can explain the selection into self-employment of both highskilled and less skilled migrants. We test the model predictions on a rich survey dataset of immigrants in Spain for 2006-2007. Our findings reject a U-shaped relationship between immigrants.skills and self-employment for the Spanish case and instead points to positive sorting into entrepreneurship. Self-employed migrants tend to have (statistically significant) better observable characteristics than salaried workers. However, non-market mechanisms, that is, penalties in the labour market beyond the mere human capital losses than migrants experience upon arrival, are also consistent with the relatively higher probability of self-employment and the lower entrepreneurial quality of certain migrant groups.