Nuclear war, beyond its devastating direct impacts, is expected to cause global climatic perturbations through injections of soot into the upper atmosphere. Reduced temperature and sunlight could drive unprecedented reductions in agricultural production, endangering global food security. However, the effects of nuclear war on marine wild-capture fisheries, which significantly contribute to the global animal protein and micronutrient supply, remain unexplored. We simulate the climatic effects of six war scenarios on fish biomass and catch globally, using a state-of-the-art Earth system model and global process-based fisheries model. We also simulate how either rapidly increased fish demand (driven by food shortages) or decreased ability to fish (due to infrastructure disruptions), would affect global catches, and test the benefits of strong prewar fisheries management. We find a decade-long negative climatic impact that intensifies with soot emissions, with global biomass and catch falling by up to 18 ± 3% and 29 ± 7% after a US–Russia war under business-as-usual fishing—similar in magnitude to the end-of-century declines under unmitigated global warming. When war occurs in an overfished state, increasing demand increases short-term (1 to 2 y) catch by at most ∼30% followed by precipitous declines of up to ∼70%, thus offsetting only a minor fraction of agricultural losses. However, effective prewar management that rebuilds fish biomass could ensure a short-term catch buffer large enough to replace ∼43 ± 35% of today’s global animal protein production. This buffering function in the event of a global food emergency adds to the many previously known economic and ecological benefits of effective and precautionary fisheries management.