Recent studies have suggested that a high level of recent physical activity increases the risk of a wrist fracture in postmenopausal women. The relationship of more distant past physical activity to wrist fracture is less clear, and most studies have relied on recall of physical activity much earlier in life. The aim of this study was to assess the risk of wrist fracture in a subset of women who had completed a recent questionnaire and also had participated in a cohort study 25 years earlier, 1865 women who were perimenopausal and postmenopausal in 1976 and had completed the 1976 and 2002 Adventist Health Study lifestyle questionnaires. Data on risk factors including physical activity were collected from the 1976 survey. Subjects reported wrist fractures occurring since baseline, and the approximate time of fracture, in the 2002 questionnaire. Incidence of wrist fracture was 3.7/1000 person-years of follow up. There was a dose-response inverse relationship between level of physical activity and wrist fracture with a 37% reduction of risk for the highest level of physical activity with respect to the lowest level (HR, 0.63; 95% CI, 0.45, 0.89). The effect of physical activity changed little in the final multivariable model (HR, 0.61; 95% CI, 0.43, 0.87). In this cohort of women with a relatively low incidence of wrist fracture, higher levels of physical activity at baseline were protective against risk of fracture during 25 years of follow-up.