Background: We examine the association between leisure-time activities and the risk of developing cognitive impairment among Chinese older people, and further investigate whether the association varies by educational level. Methods: This follow-up study included 6,586 participants (aged 79.5 9.8 years, range 65-105 years, 51.7% female) of the Chinese Longitudinal Healthy Longevity Survey who were aged 65 years and were free of cognitive impairment in 2002. Incident cognitive impairment was defined at the 2005 or 2008/2009 survey following an education-based cut-off on the adapted Chinese version of Mini-Mental State Examination (MMSE). Participation in cognitive activities (e.g. reading) and non-exercise physical activity (e.g. housework) was assessed by a self-reported scale. Cox proportional hazard models were employed to examine the association of leisure activities with incident cognitive impairment while controlling for age, gender, education, occupation, residence, physical exercise, smoking, drinking, cardiovascular diseases and risk factors, negative well-being, and physical functioning, and baseline MMSE score. Results: During a five-year follow-up, 1,448 participants developed incident cognitive impairment. Overall, a high level of participation in leisure activities was associated with a 41% decreased risk of cognitive impairment compared to low-level engagement in leisure activities after controlling for age, gender, education, and other confounders. Moreover, there was a significant interaction between leisure activity and educational level, such that the beneficial effect of leisure activities on cognitive function was larger in educated elderly than their uneducated counterparts, and only educated elderly benefited from cognitive activities. Conclusions: Late-life leisure activities protect against cognitive impairment among elderly Chinese people, and the protective effects are more profound for educated elderly.