OBJECTIVE: To investigate the reliability of self-report of menstrual frequency and flow changes in the context of determining menopausal status categories, using data from the longitudinal phase of the Melbourne Women's Midlife Health Project (MWMHP). METHODS: Women reporting at interview at least one menstrual period during the previous 3 months are assigned pre- or perimenopausal status according to their responses to questions about changes in menstrual frequency and flow. For a sample of 72 such women, menstrual diary information was converted into standardized scores measuring change in frequency and flow of menses during the 2 years prior to interview. These scores, coded into categories, were used to derive measures of the sensitivity, specificity and predictive values of the interview responses. RESULTS: Self-report of change in menstrual frequency and flow have low sensitivity to measures based on prospectively kept menstrual diaries. CONCLUSIONS: Retrospective self-report at interview of changes in menstrual frequency and flow should not be regarded as reliable measures of actual changes in cycle parameters.