Using Israeli data, we estimate the effect of school entrance age (SEA) on student outcomes. Unlike much of the recent literature, our identification strategy strictly satisfies the monotonicity assumption required for interpreting our estimates as the local average treatment effect (LATE), and also separates the effect of SEA from date of birth effects. We find that delaying school entry by one year increases fifth grade test scores in Hebrew by 0.34 standard deviations and in math by 0.19. Interestingly, while the advantage in Hebrew slightly decreases in eighth grade, in math it almost doubles. We also show that by failing to control for date of birth fixed effects we would have erroneously concluded that the SEA effect on math test scores decreases slightly from fifth grade to eight grade while it actually substantially increases.