To evaluate the relative effect of autecologic factors, site-specific factors, disturbance characteristics, and community structure on the recovery of temperate-stream fish communities, we reviewed case histories for 49 sites and recorded data on 411 recovery end points. Most data were derived from studies of low-gradient third- or fourth-order temperate streams located in forested or agricultural watersheds. Species composition, species richness, and total density all recovered within one year for over 70% of systems studied. Lotic fish communities were not resilient to press disturbances (e.g., mining, logging, channelization) in the absence of mitigation efforts (recovery time >5 to >52 yr) and in these cases recovery was limited by habitat quality. Following pulse disturbances, autecological factors, site-specific factors, and disturbance-specific factors all affected rates of recovery. Centrarchids and minnows were most resilient to disturbance, while salmonid populations were least resilient of all families considered. Species within rock-substrate/nest-spawning guilds required significantly longer time periods to either recolonize or reestablish predisturbance population densities than did species within other reproductive guilds. Recovery was enhanced by the presence of refugia but was delayed by barriers to migration, especially when source populations for recolonization were relatively distant. Median population recovery times for systems in which disturbances occurred during or immediately prior to spawning were significantly less than median recovery times for systems in which disturbances occurred immediately after spawning. There was little evidence for the influence of biotic interactions on recovery rates.