To assess the appropriateness of a single measurement of urinary 6-sulfatoxymelatonin (aMT6S) as a marker for long-term exposure to endogenous melatonin in epidemiological studies, we examined the reproducibility of aMT6S in first morning urine voids collected from 40 postmenopausal women. Urine specimens were collected on three different occasions, and the mean time between the first and the third urine sample was 5.1 years. Urinary aMT6S levels were measured by radioimmunoassay and adjusted for creatinine. The intraclass correlation for aMT6S adjusted for creatinine was 0.56 (95% confidence interval, 0.39-0.73). The classification of aMT6S concentrations in first morning voids from postmenopausal women appears to be sufficiently reproducible to justify its use as a marker for long-term exposure to melatonin in epidemiological studies.