Determinants of age at menopause were evaluated using data of control subjects collected in the framework of a case-control study on breast and female genital tract neoplasms. After exclusion of women with surgical or pharmacological menopause, a total of 863 post-menopausal women aged 55-74 years admitted to a network of general and university hospitals for non-neoplastic, non-gynecological, non-endocrine-related acute conditions were considered for this analysis. No relationship emerged between cohort of birth, marital status and mean age at menopause. No association was observed with education, but women with 12 years of schooling or more reported a slightly later mean age at menopause (50.2 years vs. 49.3 for < 7 years of schooling). There was a strong, statistically significant association with smoking, mean age at menopause being more than 1 year earlier in ever-smokers than in never-smokers. Nulliparous women tended to report an earlier age at menopause (49.0 years) than parous ones (49.5) and mean age at menopause rose with number of births, being 49.8 in women reporting three or more births. This trend was of borderline statistical significance. No association emerged between age at menopause and number of spontaneous abortions, lifelong menstrual pattern and age at menopause. The significant association between smoking, parity and age at menopause observed in the univariate analysis was confirmed after taking into account the potential confounding effects of age and education in multivariate analysis.