We compare a set of econometric studies that measure the effect of net internal migration in neoclassical models of long-run real income convergence and derive 67 comparable effect sizes. The precision-weighted estimate of beta convergence is about 2.7 per cent. An increase of one percentage point in the net migration rate of a region increases the per capita income growth rate in that region on average by about 0.1 percentage points. Introducing a net migration variable in a growth regression increases the estimate of beta convergence slightly. Studies that use panel models or IV estimation methods yield smaller coefficients of net migration in growth regressions, while the opposite holds for regressions controlling for high-skilled migration.