We investigated the associations of physical activity of elderly residents with built environment and social capital in Takashimadaira, a large residential complex in Tokyo, Japan. As built environment indices, we adopted floor number and the distance to the supermarket in front of the nearest neighbor station. Also, as physical activity indices, we adopted step counts adjusted for age, sex and month, ratio of step counts on rainy days to that on non-rainy days and ratio of step counts after/before intervention. We found that step counts of residents living far from the nearest station and on 7th floor or upper decreased more and those of residents who participated in community activities or volunteer activities decreased less on rainy days.