Previous studies have examined the relationship between body mass index (BMI) and physical fitness in older adults, but no conclusive research results have been obtained. The purpose of this study was to examine the relationship between BMI and selected physical fitness variables in older adults. We recruited 464 independently-living older adults (156 men and 308 women), aged 60-80 years, from two communities. BMI was calculated as body weight (kg) divided by the squared height (kg/m^2). Based on percentile rank of BMI, participants were categorized into three groups: (1) low BMI (men: <21.1, n = 31; women: <20.8, n = 61), (2) middle BMI (men: 21.1-26.0, n = 94; women: 20.8-26.2, n = 186), and (3) high BMI (men:≧ 26.0, n = 31; women:≧ 26.2, n = 61). Participants completed 11 physical performance tests related to activities of daily living (ADL). The 11 performance tests were summarized as three fitness domains: total physical fitness, physical fitness of upper limbs, and physical fitness of lower limbs. In addition, the 11 performance tests were summarized as four ADL domains: locomotion of whole body, manipulation of upper limb, manipulation of finger, and change of posture. These domains among the three groups were compared using ANCOVA, in which age was used as the covariate. For men, manipulation of the upper limb was found to be poorer in the low BMI group than in the other two groups, and for women was poorer in the low BMI group than in the high BMI group (P<0.05). For women, physical fitness of the lower limbs and change of posture were found to be poorer in the high BMI group than in the other two groups (P< 0.05). These results suggest that low BMI and high BMI are related to selected physical fitness parameters among older adults, and may affect the ability perform ADL, regardless of gender.