The purpose of this review was to look for trends across time in response rates and variables studied for published mail surveys and to compare response rates and variables studied for different target populations. Studies were identified in databases in four fields: education, psychology, business and marketing, and sociology. A total of 225 articles published between 1931 and the present were selected and classified according to population group and a number of survey characteristics. Mean response rates show a cyclical pattern, with the 5-year period with the highest response rate being 1956 through 1960. There was a decline for the following 10 years, with a rebound in 1971-75, followed by another drop and a rebound in 1986-90. Between 1991-95, the response rate was within one percent of the lowest mean percentage of any period studied. As journals lower their standards (response rate required) for publication, they might increase the number of articles accepted, and this may affect the mean response rate for the period. The overall response rate was affected by the disproportionate number of surveys of the general public. The study also suggests that more highly educated populations tend to have higher response rates. It was disappointing not to find a trend of increasing response rates as researchers learn more about the relative effectiveness of survey procedures. Why this should be so raises interesting research questions. (Contains 6 tables.) (SLD)