BACKGROUND: Recent findings from randomized clinical trials on the effects of hormone replacement therapy (HRT) among postmenopausal women contradict findings from observational studies indicating a protective effect on the development of cardiovascular disease. Most observational studies on HRT are based on self-reported data, although data on the validity of HRT in postmenopausal women are sparse. METHODS: We examined self-reported HRT use from questionnaires administered in 1993 (n = 2694) and again in 1999 (n = 2666) to a cohort of Danish nurses living in two Danish counties compared with prescription-reimbursement data from two administrative databases through the Danish National Health Service. RESULTS: The sensitivity and specificity of the self-reported, current HRT use in 1993 were 78.4%[95% confidence interval (95% CI) 75.4-81.4] and 98.4% (95% CI 97.8-98.9), respectively. In 1999, the estimates were 74.8% (95% CI 72.0-77.7) and 98.0% (95% CI 97.3-98.8), respectively. None of the factors examined--including age, alcohol intake, physical activity, smoking, presence of hypertension, and body mass index--was strongly associated with validity. We found a relatively high validity of self-reported data on HRT use. Furthermore, agreement between self-reported and registry-based data was not strongly associated with a range of demographic and lifestyle factors. CONCLUSION: These findings suggest that use of self-reported data is not an important contributor to the apparent discrepancy between observational studies and randomized trials on the cardiovascular effects of HRT use.