Pré-publication, Document de travail: Despite the worldwide decrease in fertility rates, Sub-Saharan Africa is still an exception, showing an almost non-declining trend over the past 50 years. In a high child mortality context parents might prefer a larger number of children, anticipating the risk of child mortality. This paper tests the short-term impact of an exogenous decrease in child mortality on household fertility. By exploiting positive exogenous weather shocks together with household panel data, I find that abundant rainfall increases child survival in the Nigerian context. Large households are the ones who benefit the most from this, and they are also the ones who respond by decreasing their fertility the most. Conversely, small households only slightly benefit from a decrease in child mortality and they continue to increase their birth rate. For a household with the average number of three children, mortality decreases by 0.013 while fertility increases by 0.046 children. When positive shocks occur, households get on average larger, as more children survive and parents only partially reduce their fertility. Consistent with such partial adjustment, household food security and children's anthropometric measures deteriorate. This matches the predictions of the theoretical framework, which shows that the magnitude of the fertility adjustment depends on the number of children alive at the moment of the shock. The empirical analysis tests this prediction, by using the gender of the first-born as instrument for the initial number of children.