Using a longitudinal survey from rural Guatemala, we examine the effect of an early childhood nutritional intervention on adult educational outcomes. An intent-to-treat model yields substantial effects of an experimental intervention that provided highly nutritious food supplements to children, a quarter century after it ended: increases of 1.2 grades completed for women and one quarter SD on standardised reading comprehension and non-verbal cognitive ability tests for both women and men. Two-stage least squares results that endogenise the actual supplement intakes corroborate these magnitudes. Improving the nutrient intakes of very young children can have substantial, long-term, educational consequences.