The study reports on factors predicting the longevity of 328 people over the age of 65 drawn from an English city and followed over 20 years. Both the reported activities score and the individual's comparative evaluation of their own level of activity independently reduced the risk of death, even when health and cognitive status were taken into account. The analysis has provided a strong test of the relevance of measures of reported activity and measures of self-perception to longevity. The study confirms the important predictive role of reported activity levels even when detailed health measures are taken into account. But in addition personal perception of one's own relative level of activity, and to a lesser extent subjective assessment of health, also predicted longevity.