Objectives. The aim of this study was to investigate whether leisure activity is associated with incident dementia in an older sample. Method. We examined a sample of 1,475 elderly (>= 65 years) who were dementia free at baseline over a follow-up period of up to 15 years. In addition to analyses involving the total time period, separate analyses of three time periods were performed, 1-5, 6-10, and 11-15 years, following baseline measurement of leisure activity. Results. After controlling for a variety of potential confounders, analyses of data for the total time period revealed that higher levels of Total activity and Social activity, but not Mental activity, were associated with decreased risk of dementia. However, analyses of the separate time periods showed that this association was only significant in the first time period, 1-5 years after baseline. Discussion. The results from this study provide little support for the hypothesis that frequent engagement in leisure activities among elderly serve to protect against dementia diseases across a longer time frame. The finding of a relationship for the first time period, 1-5 years after baseline, could indicate short-term protective effects but could also reflect reverse causality.