The purpose of this study was to test the hypothesis that an active lifestyle delays age at dementia onset. This study included 388 incident dementia cases (DSM-III-R criteria) that developed over a 9-year follow-up period among 1,375 baseline dementia-free community dwellers with good cognitive function (MMSE >23) (mean age = 81.2) from the Kungsholmen Project. An active lifestyle was defined as participation in mental, physical, or social activity. We used linear regression models to estimate influence of baseline active lifestyle on age at onset of incident dementia and general linear models to estimate mean age at dementia onset. Age at onset of dementia was significantly older in persons who had higher levels of participation in mental, physical, or social activity (β: 0.18, 0.29 and 0.23 respectively, p < 0.001 for all the activities) independent of education, medical condition, functional status, and other confounders including APOE. When the three types of activities were integrated into an index, we found that the broader the spectrum of participation in the activities, the later the onset of disease (β = 0.93, p = 0.01 for participating in two activities, and β = 1.42, p < 0.001 for three activities). There were 17 months difference in mean age at dementia onset between the inactive group and the most active group. An active lifestyle operates as a protective factor for dementia by delaying the clinical onset of the disease. These findings highlight the relevance of encouraging old adults to have active lifestyles, which could have a great impact on public health.