The relationship between reproductive variables (parity, age at first birth, number of induced and spontaneous abortions) and cancer risk has been analysed using data from an integrated series of case-control studies conducted in northern Italy between 1983 and 1992. The overall data-set included women below age 75 with histologically confirmed cancers of the following sites: oesophagus, 58; stomach, 280; colon, 405; rectum, 210; liver, 82; gall-bladder, 29; pancreas, 129; breast, 3,415; cervix, 742; endometrium, 725; ovary, 953; bladder, 68; kidney, 56; thyroid, 180; lymphomas, 80; myelomas, 57; and a total of 5,619 controls admitted to hospital for acute non-neoplastic, non-gynaecological, non-hormone-related conditions. Multivariate odds ratios, as estimators of relative risks (RR), were obtained after allowance for age, education, use of oral contraceptives and oestrogen replacement treatments, plus various reproductive factors. Direct significant trends with parity were observed for cancer of the liver (RR for women with > or = 4 births vs. nulliparae = 3.3) and cervix uteri (RR = 4.1). The risk of gall-bladder cancer was also elevated for multiparae (RR = 1.9). No significant inverse trend in risk emerged. However, the RRs in multiparae were significantly below unity for breast (RR = 0.8), endometrium (RR = 0.7), and ovary (RR = 0.8). With reference to age at first birth, a significant trend in risk was observed for breast cancer (RR = 1.4 for 25 to 29 and 1.5 for > or = 30 vs. < 25 years). In contrast, the risk of cervical cancer was inversely related to age at first birth. For spontaneous abortions, the only significant inverse trend was for ovarian cancer (RR = 0.7 for > or = 2 vs. 0 abortions), but also the point estimate for endometrial cancer in women with > or = 2 abortions was below unity. For induced abortions, there was a strong inverse trend in risk for endometrial cancer (RR = 0.5), and the RRs were below unity also for colon and breast cancer. In contrast, cervical cancer was directly associated with the number of spontaneous abortions. Although the underlying aetiological interpretations are different for various cancer sites, this study provides, in a large and uniform data-set, quantitative information on the long-term impact of reproductive factors on cancer risk.