Models for all-cause mortality among 45000 men and women with cancer in 12 different sites were estimated, using register and census data for complete Norwegian birth cohorts. This observed-survival method seemed to be an adequate approach. The results support the idea that women who were pregnant shortly before a breast cancer diagnosis may have a poorer prognosis than others. In principle, such an effect may also reflect that these women have a young child during the follow-up period, and are burdened by that. However, this social explanation can hardly be very important, given the absence of a corresponding significant effect in men and for other cancer sites in women. Breast cancer is different from other malignancies also with respect to the effect of parenthood more generally, regardless of the timing of the pregnancies. On the whole, male and female cancer patients with children experience a lower mortality than the childless, although without a special advantage associated with adult children. This suggests a social effect, perhaps operating through a link between parenthood, life style and general health. No parity effect was seen for breast cancer, however, which may signal that the social effect is set off against an adverse physiological effect of motherhood for this particular cancer. Among men, both marriage and parenthood were associated with a good prognosis. Married male cancer patients with children had a mortality 1/3 lower than that among the childless and never-married. Women who had never married did not have the same disadvantage.
