In this paper, the author estimates a multinomial logit model to analyse the contextual determinants of internal migration in Sri Lanka. The study uses micro-level data from the Consumer Finance and Socio-Economic Survey 2003/2004 of the Central Bank of Sri Lanka, and finds that higher probability for internal migration is correlated with rural areas; districts with larger concentration of labour in agriculture; districts with a larger share of population with ownership of assets; and districts with a larger share of population in 19–34?years. Districts with large labour forces and districts with a more unequal income distribution are correlated with a lower probability for internal migration. Policy recommendations of the findings include, repositioning of the balance in policy focus from receiving areas of internal migrants towards sending areas of internal migration by developing the rural infrastructure to efficiently receive remittances, improve investment of these remittances in agriculture technology to ensure agriculture productivity and focus on demographic composition of those left behind. Rural areas with larger labour forces as well as areas with greater income inequality are ideal candidates to be developed in metro cities as noted by the development policy framework Department of National Planning.