The hypothesis of the study was that social contacts to close friends and relatives and perceived social integration was able to delay mortality in general and cardiovascular mortality in particular. Altogether 1752 males and females, aged 70-100 years were interviewed by trained nurses in 1972 to 1974. The study group was based upon a random sample of all elderly in the town of Odense, Denmark. More than 80% participated in the survey which included data collection on social networks and health at the time of interviewing. By means of linking the study group to national registries on mortality and causes of mortality practically all in the cohort were traced until 1987. During follow-up 1501 persons died. Most of the association between social networks and mortality were weak and statistically insignificant but had the expected sign. After adjusting for initial health status only the interviewer's assessment of the quality of the network was statistically significant associated with longevity. A feeling of loneliness was found to be associated with cardiovascular mortality, especially for males.