This paper examines the short-run impact of internal emigration in Mexico on the physical and mental health of the migrant and the accompanying (or remaining) family members. We examine differential impacts based on whether individuals belong to a current migrant household - with member(s) at destination - or a return migrant household. We also distinguish between moves to a city and non-city. Our empirical strategy exploits the longitudinal nature of the Mexican Family Life Survey and uses a latent factor structure in which migration is specified as a multinomial choice to account for self-selection into migration based on health or other hard-to-observe characteristics. The results suggest that there are significant self-selection effects that bias the migration effects towards zero. Our estimates of short-run impacts indicate that migration to a city leads to more illnesses and depressive symptoms among adults and non-city migration lowers average weight-for-age among children. While current migration worsens both adult physical and mental health, return migration worsens adult physical health but improves mental health. Since the evidence suggest that the health effects of migration vary with the nature of migration, previous studies that treat migration as a binary decision provide an incomplete picture of the consequences of migration.