We study the effects of school starting age on siblings' infant health. In Spain, children born in December start school a year earlier than those born the following January, despite being essentially the same age. We follow a regression discontinuity design to compare the health at birth of the children of women born in January versus the previous December, using administrative, population-level data. We find small and insignificant effects on average weight at birth, but, compared to the children of December-born mothers, the children of January-born mothers are more likely to have very low birthweight. We then show that January-born women have the same educational attainment and the same partnership dynamics as December-born women. However, they finish school later and are (several months) older when they have their first child. Our results suggest that maternal age is a plausible mechanism behind our estimated impacts of school starting age on infant health.