AN the last decade there has been increasing concern over response rates in sample surveys of the United States household population. Reports of major declines in survey response and criticisms of survey methods led the American Statistical Association in 1973 to sponsor a Conference on Surveys of Human Populations. Lamenting the unavailability of adequate data-particularly from research organizations operating in university settings-the conference concluded that achieving satisfactory completion rates had become more difficult. At the same time it recognized that disagreement existed within the survey research community about the possible causal factors. Opinion was split between those who felt the problem stemmed primarily from an increase in the numbers of respondents who could not be found at home and those who believed the change was due to an increase in refusals. In any case, the conference called for a "more intensive examination of the problem" and recommended the establishment of a full-time research staff whose main task would be Abstract Reported increases in nonresponse rates to sample surveys have not been systematically documented to date. Data from the National Election Studies and the Surveys of Consumer Attitudes, two well-known continuing studies conducted by the Survey Research Center at the University of Michigan, permit the assessment of long-term trends for the two major components of nonresponse, refusals and other noninterviews, by urban subgroups using time series regression techniques. The analyses clearly demonstrate that there have been substantial increases in total nonresponse due primarily to increases in the percentages of respondents who refuse to be interviewed, and that these trends are related to the level of urbanization.