A random sample of non-institutionalised Swedish elderly (n = 1,062; aged 67+) were interviewed in 1954. All of them are now deceased; their interview data have been completed with dates of death and causes of death. A measure of survival capacity has been used, based on the endured total mortality risk from examination until death, according to life tables from Statistics Sweden. Survival analyses were performed by gender on the whole sample and on a number of sub-samples defined by age, health-status, social class, and marital status at examination. Physical health status and activity patterns had the overall most significant effects on subsequent longevity. Among the youngest elderly mental health was, however, even more important as a predictor of survival than was physical health. The impact of most other factors like social networks, mobility, and religiosity was of less magnitude, but their importance varied among sub-samples. Most of the variation in survival, however, remains unexplained.