To investigate possible correlates of the systematically higher pancreatic cancer rates in males than in females, the role of menstrual, reproductive and hormonal factors in females have been assessed using data from a case-control study conducted in Northern Italy. Cases were 133 women with histologically confirmed incident cancer of the pancreas, and controls were 377 women in hospital for acute, non-neoplastic, non-digestive-tract disorders. After allowance for age, education, area of residence and smoking habit, an increased risk of pancreatic cancer was observed in women with early menarche (< or = 13 years) (OR = 1.9; 95% CI: 1.0-3.6), but no significant association was observed with age at menopause or length of fertile life. Parous women were at reduced risk as compared to nulliparous women (OR = 0.7), although the trend in risk with number of births was not significant. No association with spontaneous or induced abortions was observed. Pancreatic cancer risk was inversely related to early age at first birth (first birth < 25 versus nulliparae: OR = 0.5; 95% CI: 0.3-0.9; p-value for trend < 0.01) and to age at last birth (last birth < 25 versus nulliparae: OR = 0.3; 95% CI: 0.1-0.8; p-value for trend < 0.05). Ever-users of estrogen replacement therapy showed a non-significantly increased risk (OR = 2.2). Although no clear pattern of association is evident, the present results are in agreement with previous epidemiological observations and experimental research indicating that hormonal (menstrual and reproductive) factors could explain part of the male-to-female differential in incidence and mortality from pancreatic cancer.