Rigorously designed social surveys, not public complaint actions, provide the most direct available evidence about the impact of environmental noise on residents. The balance of the available social survey evidence indicates that while personal attitudes affect noise annoyance, demographic characteristics and ambient noise levels do not affect annoyance with audible sounds. A continuously graded annoyance reaction to noise does not provide a strong scientific basis for choosing ‘‘highly annoyed’’ rather than any other degree of annoyance as an acceptability criterion. A rigorous synthesis of annoyance survey results must screen out erroneous data, use objective methods for selecting data sets, systematically adjust for important differences in nominal noise measurement conditions, weight data points by their precision, have a firm empirical basis for relating diverse annoyance measures, and satisfactorily evaluate the likely precision of the resulting synthesis curve. A theoretically sound justification for the form of a dichotomous dose/response curve must include a theory about the distribution of measured reactions to noise. Existing dose/response curves continue to make useful contributions to noise policy. Additional work will be needed before a synthesis of community dose/response relationships can be developed that meets the previously mentioned criteria.