Catastrophic mortality events that drastically reduce the abundance of a population or a particular life stage can have long-term ecological and economic effects, and are of great concern in species conservation and management. Severe die-offs may be caused by natural catastrophes such as disease outbreaks and extreme climates, or human-caused disturbances such as toxic spills. Forecasting potential impacts of such disturbances is difficult and highly uncertain due to unknown future conditions, including population status and environmental conditions at the time of impact. Here, we present a framework for quantifying the range of potential, population-level effects of catastrophic events based on a hindcasting approach. A dynamic population model with Bayesian parameter estimation is used to simulate the impact of severe (50-99%) mortality events during the early life stages of Northeast Arctic cod (Gadus morhua), an abundant marine fish population of high economic value. We quantify the impact of such die-offs in terms of subsequent changes in population biomass and harvest through direct comparison of simulated and historical trends, and estimate the duration of the impact as a measure of population resilience. Our results demonstrate strong resilience to catastrophic events that affect early life stages owing to density dependence in survival and a broad population age structure. Yet, while population recovery is. relatively fast, losses in harvest and economic value can be substantial. Future research efforts should focus on long-term and indirect effects via food web interactions in order to better understand the ecological and economic ramifications of catastrophic mortality events.