We explore how early-life shocks interact with subsequent human capital investments to influence children’s long-term outcomes. Using large-scale administrative data from Colombia, we combine a difference-in-difference framework with a regression discontinuity design to exploit two sources of exogenous variation: i) early-life exposure to adverse weather shocks that affect children’s initial skills and ii), the introduction of conditional cash transfers (CCT) that promote investments in children’s health and education. We show that the timing and type of CCT-induced investments matter for both the effects of CCTs and their interactive effects with weather shocks. When the CCT-induced investments occur in sensitive periods of human capital formation (e.g., early childhood), the effects are large and their interactive effects with weather conditions suggest that the returns of the program are even larger for children exposed to “normal” weather conditions. In contrast, CCT-induced investments that come relatively late in childhood (e.g., adolescence), have a smaller “main” effect and a smaller or zero interactive effect with weather shocks. We also find that initial CCT-induced health investments tend to have larger returns than initial CCT-induced educational investments. These findings shed new light on the developmental production function for human capital and the role of social policies in closing gaps generated by early-life adversities.