The standard human capital model of labor migration has been successful in explaining several empirical regularities of the migration process. However, there are a number of empirical facts that have not yet been explained or empirically tested; among them, that migrants usually perform more than one move in their lifetime; return migration accounts for a very important share of total migration rates; and there is a high positive correlation between in- and out-migration rates in the more advantageous regions. This paper presents both a simple model explaining these facts and an empirical analysis using data from Peru.