Background: We investigated the association between positive affect and incident dementia among the old. Studies have reported the role positive affect has on maintaining health. Still, no longitudinal studies have assessed the association between positive affect and incident dementia. Methods: We used the Aichi Gerontological Evaluation Study (AGES) data. Participants were older adults (65 +) who did not receive benefits from Japan's public Long-Term Care Insurance System at baseline (N = 14,286) in 6 municipalities. They were followed from 2003 to 2007 for dementia onset. Dementia onset was determined according to the criteria used in the Long-Term Care Insurance System. Positive affect was assessed by sub-scales of the Geriatric Depression Scale (GDS-15). Cox hazard proportional models stratified by sex were employed to calculate hazard ratios for incident dementia. Results: Of 14,286 participants (6,813 men and 7,473 women), 333 men (4.9%) and 468 women (6.3%) developed dementia during the 4 year follow-up. In age adjusted Cox models, positive affect was significantly associated with lower risk of dementia both among men and women. Even after adjusting for health status, health behaviors, social engagement, and low education, positive affect persisted as a significant protector against dementia. Conclusions: We observed a protective role of positive affect against cognitive decline. Factors associated with higher positive affect scores were healthier life style, social engagement, and physical health. This implies the importance of maintaining such activities to promote cognitive health among the old. In doing so, the role of positive affect merits attention.