OBJECTIVES: To assess if tai chi, a traditional Chinese form of exercise, could improve proprioception in old people and if the effects of tai chi on proprioception are more evident than other exercise forms in the elderly. METHODS: By detecting the threshold of passive movement, ankle and knee joint kinaesthesis was measured in 21 elderly long term tai chi practitioners (TC group), 20 elderly long term swimmers/runners (S/R group), and 27 elderly sedentary controls (control group). RESULTS: Ankle joint kinaesthesis differed significantly among the three groups (p = 0.001). Subjects in the TC group could detect a significantly smaller amount of motion than those in the S/R group (p = 0.022) and control group (p = 0.001). No significant difference was found between the S/R group and the control group (p = 0.701). The threshold for detection of passive motion was significantly different in knee extension and flexion. For knee flexion, the TC group showed a significantly lower mean threshold for detection of passive motion than the control group (p = 0.026). There were no significant differences between the S/R group and control group (p = 0.312), or between the TC group and S/R group (p = 0.533). For knee extension, no significant difference was noted among the three groups (p = 0.597). CONCLUSIONS: The elderly people who regularly practiced tai chi not only showed better proprioception at the ankle and knee joints than sedentary controls, but also better ankle kinaesthesis than swimmers/runners. The large benefits of tai chi exercise on proprioception may result in the maintenance of balance control in older people.