Taking the example of Burkina Faso, where massive rural exodus continues to drive the urbanization process, this study tests the net effects of migration on urban-rural mortality differentials among adults aged 15-74. It is based on information collected at two demographic surveillance sites, one in a rural area (Nouna) and one in the city (Ouagadougou). The longitudinal data cover a recent period (2009-2013) and are analysed using a semi-parametric Cox model. In a rural environment with poor health conditions, it is the healthy individuals who migrate to the city, thereby strengthening the urban health advantage over the short term. While rural-urban migrants are positively selected in terms of health, they tend to lose their advantage over time as they adapt to life in the city. This deterioration after several years of urban residence is probably holding back the mortality transition at national level, given that these migrants were in better health in their environment of origin. For return migrants (rural-urban-rural), the absence of a negative selection effect in Burkina Faso again reflects the complex interplay between migration and health. Beyond compositional and contextual effects, the positive selection of rural-urban migrants is accentuating the health disadvantage of rural areas.