Virtually all of the variance in data on the extent of urban noise‐induced annoyance in 24 American communities was predictable by stepwise regression from speech‐interference information. Both the annoyance and speech‐interference data were collected in the course of a National Urban Noise Survey, an EPA‐sponsored study intended to examine response to non‐aircraft, nonhighway noise in communities of widely varying exposure levels, population densities, and geographic areas. Neither complaint rates nor demographic factors alone were found to be usefully predictive of the extent of community annoyance.