Population cycles of herbivores are thought to be driven by trophic interaction mechanisms, either between food plant and herbivore or between the herbivorous prey and its natural enemies. Observational data have indicated that hymenopteran parasitoids cause delayed density-dependent mortality in cyclic autumnal moth (Epirrita autumnata) populations. We experimentally tested the parasitism hypothesis of moth population cycles by establishing a four-year parasitoid-exclusion experiment, with parasitoid-proof exclosures, parasitoid-permeable exclosures, and control plots. The exclusion of parasitoids led to high autumnal moth abundances, while the declining abundance in both the parasitoid-permeable exclosures and the control plots paralleled the naturally declining density in the study area and could be explained by high rates of parasitism. Our results provide firm experimental support for the hypothesis that hymenopteran parasitoids have a causal relationship with the delayed density-dependent component required in the generation of autumnal moth population cycles.