Partial migration is a common and widespread phenomenon in animal populations. Even though the ecological causes for the evolution and maintenance of partial migration have been widely discussed, the consequences of the genetics underlying differences in migration patterns have been little acknowledged. Here, I revise current ideas on the genetics of partial migration and identify open questions, focussing on migration in birds. The threshold model of migration describing the inheritance and phenotypic expression of migratory behaviour is strongly supported by experimental results. As a consequence of migration being a threshold trait, high levels of genetic variation can be preserved, even under strong directional selection. This is partly due to strong environmental canalization. This cryptic genetic variation may explain rapid de novo evolution of migratory behaviour in resident populations and the high prevalence of partial migration in animal populations. To date the threshold model of migration has been tested only under laboratory conditions. For obtaining a more realistic representation of migratory behaviour in the wild, the simple threshold model needs to be extended by considering that the threshold of migration or the liability may be modified by environmental effects. This environmental threshold model is valid for both facultative and obligate migration movements, and identifies genetic accommodation as an important process underlying evolutionary change in migration status. Future research should aim at identifying the major environmental variables modifying migration propensity and at determining reaction norms of the threshold and liability across variation in these variables.