Higher prostate cancer mortality rates among widowered than married men of the same age groups plus the rarity of prostate cancer in men below age 40 years suggested a need for focusing epidemiologic studies on the late middle-age period. Duration of widowerhood and cancer deaths in spouses were studied in 169 widowers dying with prostate cancer, 451 matched control widowers dying of any causes except prostate cancer, and 150 matched control widowers dying of other types of cancer. The average duration of widowerhood and frequency of multiple wives were the same for patients with prostate cancer and controls. Predeceased wives (i.e., wives dying before their husbands) of the prostate cancer patients had breast cancer 9.9% of the time, as compared to predeceased wives of the controls, who had breast cancer 5.8% of the time. The respective endometrial cancer rates for wives of patients and controls were 2.0 and 0.2%. Neither these results nor results from a confirmatory study of cancer among wives of married prostate cancer patients were statistically significant. Overstatement of widowerhood status on 6.1% of death certificates, lack of case-control differences in durations of widowerhood, and decreasing widower-married mortality ratios as age-specific groupings were made smaller indicated that the high rates of prostate cancer among widowers may be due to artifacts of classifying and tabulating vital records.