This thesis aims to describe the participation in leisure activities of elderly subjects and to detect their possible effect, if any, on the development of dementia. The data were derived from the Kungsholmen Project, which is a community-based prospective study on aging and dementia in people aged over 74 years, living in Stockholm, Sweden. The major findings are summarized below. Study I. The pattern of participation in leisure activities was related to contextual factors as well as to mental and physical health conditions. In spite of the advanced age, the majority of the population was active, as 70% participated in at least one activity. Reading (19%) was the most prevalent individual activity, and mental activities (43%) the most prevalent activity type. Older age, female gender, low education, poor or limited social network, mental disorders, and physical limitation were all correlated with a decreased engagement in at least one activity . Contextual factors and health-related factors were differentially associated with the five activity types. Study II. We aimed to verify the hypothesis that mental, social and physical components are relevant protective factors against dementia, and that a combined beneficial effect may be present over a 6-year follow-up period. Multi-adjusted relative risks (RRs) of dementia for subjects with higher mental, physical and social component scores were 0.71 (95% CI: 0.49 1.03), 0.61 (95% CI: 0.42 0.87) and 0.68 (95% CI: 0.47 0.99), respectively. The most beneficial effect was present for subjects with high scores in all or in two of the components (RR of dementia=0.53; 95% CI: 0.36 0.78). Study III. The hypothesis that an active lifestyle may protect against dementia development was further tested by using principal component analysis to characterize the exposure. Among a set of lifestyle variables, three underlying factors were identified: physical, mental and social factors. All the factors showed an independent protective effect on dementia development. The relative risks (RRs) ranged from 0.60 to 0.70. When these factors were integrated into an Active Lifestyle Index, a significant dose-response association was observed, Compared with low level of engagement (low in at least two of the factors), the RR of dementia was 0.66 (95% CI: 0.49-0.89) for the moderate level (high scores in two factors), and RR=0.51 (95% CI: 0.31-0.85) for the high level (high scores in all three factors). Study IV. The hypothesis that an active lifestyle delays the dementia onset was verified over a 9-year follow-up period. The lifestyle factors of the mental, social and physical component scores estimated in a previous study (study II) were studied in relation to age at dementia onset. Results showed that dementia developed at a significantly later age in individuals who had a higher level of participation in activities with high physical, mental or social components. When the three components were integrated, we found that the broader spectrum of participation (higher levels in at least two of the components) the later the age at dementia onset (β 0.62; P<0.01). Conclusions. Even in the advanced age, elderly persons are still active, being limited in their participation only by mental disorders or by physical limitation. An active lifestyle, defined as a higher level of participation in leisure activities with either mental, social or physical component may decrease the risk of dementia and postpone its onset.