We have examined whether the risk of cancer among divorced or widowed Norwegian women born between 1935 and 1954 was any different from that of married women. Among a total of approximately 600,000 women, we applied a nested case-control design. Thus, the study was population-based and included 17,235 incident cases of cancer with 34,460 age-matched controls. For widowed women, there was no overall relation with cancer. For divorced women, 2 strikingly different associations were apparent. A reduced risk was seen for cancers of a number of sites, including thyroid, endometrium, colorectum, and breast, as well as malignant melanoma and hematologic malignancies, with statistically significant estimates of relative risk ranging from 0.64 to 0.84. In contrast, divorced women had a strongly elevated risk of lung and cervical cancer. Moreover, there was a gradual reduction in the relative risk of cancer at some sites with increasing age at divorce, and with duration of marriage prior to divorce. In this study of middle-aged women, the risk of cancer among widows was no different from that of married women. Divorced women had an increased risk of cancers which are related to cigarette smoking but, simultaneously, a reduced risk of cancer at a number of other sites. Since the negative associations for some cancers were strongly related to increasing age at divorce and to duration of marriage, the results may indicate that the reduction in risk is related to factors which characterize the marital period preceding divorce.