Data collected from 2,197 white ovarian cancer patients and 8,893 white controls in 12 US case-control studies conducted in the period 1956-1986 were used to evaluate the relation of invasive epithelial ovarian cancer to reproductive and menstrual characteristics, exogenous estrogen use, and prior pelvic surgeries. Clear trends of decreasing risk were evident with increasing number of pregnancies (regardless of outcome) and increasing duration of breast feeding and oral contraceptive use. Ovarian dysfunction leading to both infertility and malignancy is an unlikely explanation for these trends for several reasons: 1) The trends were evident even among the highly parous; 2) risk among nulliparous women did not vary by marital status or gravidity; and 3) risk among ever-married women showed little relation to length of longest pregnancy attempt or history of clinically diagnosed infertility. Risk was increased among women who had used fertility drugs and among women with long total duration of premenopausal sexual activity without birth control; these associations were particularly strong among the nulligravid. No consistent trends in risk were seen with age at menarche, age at menopause, or duration of estrogen replacement therapy. A history of tubal ligation or of hysterectomy with ovarian conservation was associated with reduced ovarian cancer risk. These observations suggest that pregnancy, breast feeding, and oral contraceptive use induce biological changes that protect against ovarian malignancy, that, at most, a small fraction of the excess ovarian cancer risk among nulliparous women is due to infertility, and that any increased risk associated with infertility may be due to the use of fertility drugs.