This paper investigates the linkages between the environment, income and migration, exploring the role that environmental factors play in inducing migration within a rural developing economy. The environment has a deep economic value, representing a source of living for most developing countries: exogenous shocks on the environment, determined for instance by extreme climatic events, and the gradual degradation of natural resources, associated with population pressure and human activities might induce human displacement and migration. Because of the deterioration of the environment people could be forced to seek for new opportunities or might choose to look for better opportunities, migrating either temporarily or permanently. The paper introduces a theoretical framework of analysis drawing upon the traditional economics literature on migration. Based on the theoretical framework proposed, this paper describes the specification and the estimation of the econometric model, which falls into the class of switching regression models with endogenous switching. The econometric model allows for the estimation of a migration decision function in two stages, relating migration to income and the environment, and accounting for the self-selection of migrants. The model in fact consists of two income equations, which explain respectively income for migrants and for non-migrants, as well as one criterion function, which describes the dichotomous decision to migrate. The econometric model is applied to a data set on Pakistan, containing micro cross-country data on migrants within Pakistan, as well as detailed climate and environmental variables. Some descriptive statistics and policy analysis on migration and income variables is provided, and results from the econometric estimation of the empirical model are illustrated. Non-income variables seem to play the most relevant influence on the decision to migrate, and environmental factors seem to be the key factors in influencing mobility. The policy implications of the main findings of this research are illustrated.