Ecology seeks to understand the abundance and distribution of species. Ecologists typically collect either long-time series without knowledge of the drivers or short-term experimental results, which may not scale up. We surveyed the annual population dynamics of Arctia virginalis for 40 years and conducted experiments examining the relative roles of abiotic conditions, host plants, predation, parasitoids, and viral infection. Rather than finding a single limiting factor, these factors were all important at some times or places. Annual densities varied by 1000-fold and showed evidence of a regime shift around 2002, coincident with changing precipitation. Wet sites and wet years supported higher densities, and precipitation interacted with most of the factors considered. Population control was context-dependent, but water availability was generally the relevant context. Precipitation appears to be important for other Lepidoptera in western North America. Studies that include experimental tests of population drivers are required to effectively manage insect populations.