Climate change modelers predict increasingly frequent "extreme events," so it is critical to quantify whether organismal responses (such as reproductive output) measured over the range of usual climatic conditions can predict responses under more extreme conditions. In a 20-year field study on water pythons (Liasis fuscus), we quantified the effects of climatically driven annual variation in food supply on demographic traits of female pythons (feeding rate, body size, body mass, and reproductive output). Reaction norms linking food supply to feeding rates and residual body mass were broadly linear, whereas norms linking food supply to female body size became curvilinear when a dramatic (flooding-induced) famine reduced the mean body size at sexual maturity. Thus, the reaction norms recorded over 16 years of mal" (albeit highly variable) climatic conditions gave little insight into the population's response to a more extreme nutritional crisis.