Collapses of insect pest outbreaks are often attributed to delayed density dependence of predation. The forest tent caterpillar, Malacosoma disstria Hübner (Lepidoptera: Lasiocampidae), is an outbreaking pest species defoliating mixed wood boreal forests in eastern Canada. This species presents periodic population dynamics with peaks every 10 years and outbreaks lasting 1–3 years impacting the health of its host tree. We asked about the relative importance of various sources of early-instar mortality during and after the crash of an outbreak, testing for density dependence. We used a triad set-up of complete predator exclusion, partial exclusion, and free colonies to distinguish between intrinsic mortality (caused by, e.g., pathogens) and mortality caused by flighted and walking natural enemies, and compared defoliated outbreak sites with control sites. Overall, survival was lowest in the free colonies and increased with partial and complete predator exclusion. Survival was also higher in control than in outbreak sites and higher in the final year of the outbreak than in the following year. We observed no changes in mortality from walking enemies, but an increase in intrinsic mortality and mortality from flighted enemies. in the year following the outbreak. This increase is consistent with density dependence of these mortality sources, but its occurrence in the control sites as well was unexpected. These findings show that mortality from flighted natural enemies and intrinsic source increases after the outbreak, but do not contribute to the population crash. However, they help maintain low endemic levels between outbreaks.