Disease shapes community composition by removing species with strong interactions. To test whether the absence of keystone predation due to disease produced changes to the species composition of rocky intertidal communities, we leverage a natural experiment involving mass mortality of the keystone predator Pisaster ochraceus from Sea Star Wasting Syndrome. Over four years, we measured dimensions of mussel beds, sizes of Mytilus californianus, mussel recruitment, and species composition on vertical rock walls at six rocky intertidal sites on the central California coast. We also assessed the relationship between changes in mussel cover and changes in sea star density across 33 sites along the North American Pacific coast using data from long-term monitoring. After four years, the lower boundary of the central California mussel beds shifted downward toward the water 18.7 ± 15.8 cm (SD) on the rock and 11.7 ± 11.0 cm in elevation, while the upper boundary remained unchanged. In central California, downward expansion and total area of the mussel bed were positively correlated with mussel recruitment but were not correlated with pre-disease sea star density or biomass. At a multi-region scale, changes in mussel percent cover were positively correlated with pre-disease sea star densities but not change in densities. Species composition of primary substrate holders and epibionts below the mussel bed remained similar across years. Extirpation of the community below the bed did not occur. Instead, this community became limited to a smaller spatial extent while the mussel bed expanded.