Background. Previous research has shown an increased risk of mortality after spousal bereavement, with the highest risk in the first weeks or months closest to the loss. One difficult issue in these designs is appropriate covariates and control groups.

Method. This study is based on 1993 pairs of twins discordant for marital status and on 35860 married individuals from the Swedish Twin Registry born between 1886 and 1958 and followed for marital and vital status between 1981 and 1993.

Results. Spousal bereavement was a risk factor for mortality for both men and women using the still married co-twin as a control to the widowed proband, and controlling for earlier health status and health-related risk factors. The mortality risk was higher for young-old (under 70 years) individuals, and for recently widowed than for longer-term widowed. Young-old women had a pattern with increased mortality risk during the first years after bereavement, but also a markedly decreased risk if they survived 4 years after bereavement, as compared to married women.

Conclusions. The results support a causal effect of bereavement on mortality. The decrease in risk for long-term young-old women is congruent with reports by widows of psychological growth after bereavement, involving increased sense of mastery and competence after learning to live in new sets of circumstances following the loss of their husband.