Much of the global evidence on inter-generational income mobility is based on sub-optimal data.In particular, two-stage techniques are widely used to impute parental incomes for analyses of developing countries and for estimating long-run trends across multiple generations and historical periods. We propose a machine learning method that may improve the reliability and comparability of such estimates. Our approach minimizes the out-of-sample prediction error in the parental income imputation, which provides an objective criterion for choosing across different specifications of the first-stage equation. We apply the method to data from the United States and South Africa to show that under common conditions it can limit the bias generally associated to mobility estimates based on imputed parental income.