OBJECTIVE To describe stability and change in functional ability among non-institutionalized old people in relation to sex, age, and household composition during two subsequent four-year observation periods. The study describes stability and change in total functional ability as well as in specific daily activities. METHODS The study population included a random sample of non-institutionalized 70+-year-olds studied in 1986 with follow-up in 1990 and 1995. The number of participants in the three surveys were 1,231 (participation rate 67%), 911 (90% of 1,008 survivors), and 542 (77% of 706 survivors). Functional ability was measured at all three surveys. There is a considerable health selection from survey 1 to 2 to 3 because of a high mortality among participants with poor health. RESULTS The proportions with unchanged, improved, and deteriorated functional ability were 51%, 13%, and 37% during the first four years and 50%, 9%, and 42% during the following four years. These changes were not related to age and sex. Major variations were seen with regard to change in individual daily activities with most deterioration in mobility and the more outgoing IADL activities (Instrumental Activities of Daily Living). Relatively large proportions of elderly people demonstrated at the same time improvements in some functions and deterioration in others. CONCLUSIONS A notable minority of the population improved in functional ability and a large proportion was stable over time. These observations suggest a potential for health promotion among elderly persons, which up to now has been more or less neglected in health policy.