There have been relatively few epidemiological studies to verify the information obtained from study participants on the use of menopausal hormone replacement therapy. We conducted this study as part of a case-control study of diet, hormone use, and endometrial cancer in Toronto, Ontario, Canada, 1994-1998. We compared records from 653 subjects, 30-79 years of age, with reports from their physicians on ever/never use of hormone replacement therapy and duration, type, and dose of hormone replacement therapy. A total of 88% of the interview records were in agreement with physician reports for ever/never use of hormone replacement therapy. The overall kappa value for ever/never use agreement was 0.76 (range = 0.71-0.81), and the intraclass correlation coefficient was 0.64 (range = 0.57-0.70) for duration of hormone replacement therapy use, indicating good agreement; similar correlations were seen among cases and controls for overall use, as well as estrogen- or progestogen-alone use. Concordance for brand codes was observed for about 43% of the subjects. This study suggests that information obtained by interview in case-control studies provides a reasonable measure of ever use of hormone replacement therapy and duration of use. Interviews, however, do not represent a reliable source of information on brands and dosage of hormone replacement therapy preparations.