Fertility is a main driver and outcome of long-term growth. Yet, fertility may not only interact with the level of income but also with its volatility. In pre-modern economies where formal social security was largely absent, fertility decisions may also have been made in view of insuring income shocks which were hard to predict such as an income shock due to crop disease, a shortage in rainfall, a case of serious illness or a job loss. In this paper, I focus on the demographic transition in the United States covering the period 1870 to 1930 and explore whether variation in fertility among American settlers can be explained by variations in exposure to rainfall risks. Identification relies on fertility differences between farm and non-farm households within counties and over time. The results suggest that increased rainfall risk does indeed increase fertility among farming households but not among households with other occupations that are less dependent on rainfall. The channel is robust to other relevant forces such as returns to education and children's survival. The analysis also shows that this effect is reduced if risk management devices such as irrigation systems, formal financial institutions or machinery emerge. The findings contribute to the understanding of the demographic transition in the US and in risk-prone areas more generally.