Increasing parity is associated with a reduction in the risk of ovarian cancer, but it is not clear whether this association applies to different histopathological types and to borderline tumours. Moreover, the temporal relations are poorly understood, and the possible role of age at first birth remains unequivocal. We have investigated these issues in a case-control study nested in a nationwide cohort of women born between 1925 and 1960 in Sweden. During follow-up until 1984, 3486 invasive ovarian cancers (2992 epithelial, 330 stromal, 149 germ-cell, 15 not classifiable) and 510 tumours of borderline malignant potential were diagnosed. 5 individually age-matched controls (total 19,980) were selected for each case woman. After simultaneous adjustment for parity and age at first birth, increasing parity was associated with a pronounced consistent decrease in relative risk of all invasive cancers (odds ratio for each additional birth 0.81 [95% Cl 0.77-0.85]), epithelial cancer (0.81 [0.77-0.86]), stromal cancer (0.84 [0.72-0.98]), and germ-cell cancer (0.71 [0.48-1.05]), but a less consistent decrease for borderline tumours (0.92 [0.81-1.04]). The risk of ovarian cancer decreased by about 10% for each 5-year increment in age at first childbirth (odds ratios 0.89 [0.84-0.94] epithelial cancer, 0.92 [0.77-1.10] stromal cancer, 0.92 [0.65-1.32] germ-cell cancer, 0.93 [0.80-1.09] borderline tumours). Because our findings cannot be readily explained by theories involving incessant ovulation or high serum concentrations of gonadotropins, new aetiological hypotheses are needed. Pregnancy-dependent clearance from the ovaries of cells that have undergone malignant transformation could explain the reproductive risk factors for ovarian cancer.