Education is expected to play a major role in explaining the gains from migration because it may be more valuable at the destination. This paper constructs a simple rural-to-urbanmigration model. In the model, urban jobs reward migrants for their education, and the higherreturn to schooling gives rural agents an incentive to migrate, encouraging the less educatedto invest in further schooling before joining labour force, conditional on the initially endowed level of education. Depending upon where learning opportunities are available, some migrate to seek more education, and others migrate after obtaining it and the investment process ofmigration and schooling eventually leads migrants into higher earnings. The model has im-portant implications for an empirical analysis. If the migration effect on an outcome (earningsor consumption) is entirely attributed to the higher return to schooling at the destination,the effect will disappear by controlling for the current level of education and allowing for thedifference in the return between the origin and the destination in the outcome equation. Thissuggests that those who cannot afford the cost of schooling (as well as the cost of migration)cannot reap the benefits of migration. On the other hand, if the migration effect remains inthe presence of a full set of controls, then, the evidence points to the migration effect operat-ing other than through education. Using long-run household panel data from rural Tanzania,Kagera from 1991 to 2004, this paper investigates the extent to which education can explainmigrants’ income and consumption gains by estimating those outcome equations. Empiricaldifficulty arises from the endogeneity of migration and schooling decisions. To address the en-dogeneity, this study takes two approaches: an IV approach (structural modelling approach)and a control function (CF) approach. Controlling for initial household fixed effects, we findthat individuals are positively selected into both migration and further schooling based ontheir pre-migration level of education. In consumption, we provide evidence that the return toschooling is indeed higher at the destination, but that a large significant migration effect re-mains after controlling for education and allowing for the endogenous selection into migrationand schooling. Evidence on earnings is rather ambiguous due to the difficulty in controllingfor the endogeneity, but the analysis still provides some evidence to reveal a similar story toconsumption. These suggest that income and consumption opportunities are higher at thedestination, and that some mechanisms, operating other than through education, prevent in-dividuals from exploiting the arbitrage opportunity. While education plays a role, it does notappear to be a major factor in limiting the internal migration as a source of raising incomeand consumptio