The study was to examine the longitudinal association between physical activity and subsequent severe cognitive impairment of older adult participants from a national sample. Data from Health and Retirement Study (HRS), Wave 4 (1998) and Wave 5 (2000) was used with 14,931 individuals aged 50 years or over as the final sample. Cognitive function was measured by six subscales developed by Herzog and Wallace (1997), including immediate free recall, delayed free recall, serial 7s to test working memory, and test of knowledge, language, and orientation. A score of eight or lower out of 35 total score was considered as severely cognitive impaired. Multivariate analyses were employed to adjust the confounding factors. Controlled for covariates, such as economic status, chronic condition, body composition, self-rate heath status, and social support, a binary logistic regression was used to determine the association between change of physical activity participation (from Wave 4 to Wave 5) and onset of severe cognitive impairment (Wave 5) for those who were not severely cognitive impaired at baseline (Wave 4). Significant effects of change in physical activity level on subsequent onset of severe cognitive impairment in 2000 were observed among non-impaired individuals in 1998. The Odds Ratios (OR) estimation for Remain Active (OR = 0.723) suggested that the expected odds of being severely cognitive impaired decreased by about 28% for those remaining physically active compared to the participants who became physically inactive within a two year period. The OR estimation for Become Active (OR = 0.552) suggested that the expected odds of being severely cognitive impaired decrease by about 45% for those becoming physically active comparing to those who become physically inactive within the two year period. Interestingly, no significant difference in risk of severe cognitive impairment in 2000 between the participants who were inactive versus the participants who became inactive since 1998 was existed. The elderly who remain physically active in two years are less likely to suffer from severe cognitive impairment. The elderly who were inactive and become active in two years are even less likely to suffer from severe cognitive impairment. The findings were consistent with previous experimental studies in clinical settings which have further strengthened the evidence of the protective effect of physical activity participation against severe cognitive impairment among older adults.Keyword(s): aging/older adult issues, exercise/fitness/physical activity, health promotion