The first mail reinterview ever conducted by the Census Bureau was conducted as part of the Schools and Staffing Survey (SASS) of 1991. Because there were some questions about the reliability of items from the 1988 SASS reinterview, the 1991 study was examined by a mail reinterview. The methodology and responses are discussed, and response variance rates for questions reinterviewed in the 1988 and 1991 SASS cycles were compared. Mail reinterview respondents displayed lower simple response variance than under the telephone-telephone procedure. Findings suggest that mail reinterview respondents are cooperative and careful, although they may leave more difficult or uncertain questions blank. Implications of these findings are discussed. Mail surveys may provide data that are as good or better than some surveys now conducted by telephone or in person. Future survey plans are reviewed. Six tables present study findings. (Contains 7 references.) (SLD)