Many studies of the relation between factors earlier in life and late-life cognition have a short follow-up time, often less than 10 years. Since cognitive decline can be present up to 20 years prior to dementia it is difficult to distinguish the direction of the relationships without a long follow-up. We analyzed the association between different types of leisure time activity at baseline and cognition more than 20 years later. A wide range of activities was included-political, mental, socio-cultural, social, physical, and organizational activities. Baseline studies were random Swedish samples aged 46-75 years (mean 57.4) (n=1643) interviewed in 1968 or 1981. Activities were measured at baseline. Cognition was measured with items from the Mini-Mental State Examination in 1992, 2002, or 2004. There was a significant association between later cognition and earlier political, mental, and socio-cultural activities controlling for age, age-square, sex, follow-up-time, mobility problems, symptoms of mental distress, employment status, education, adult and childhood socioeconomic status, income, smoking, and drinking. Physical activities had a significant association with cognition only among women. Organizational activities were not significant when controlling for all covariates. Social activities had no significant association. Including all covariates and all leisure activities simultaneously, only mid-life political and mental activities remained significantly related to later life cognition. These findings add support to the theory that various forms of engagement in mid-life can have a protective effect with respect to cognition in later life. Copyright © 2010 John Wiley & Sons, Ltd.