The effect of short-term training and detraining of rapid tapping tasks on motor function of the fingers was investigated in 7 healthy elderly adults. The experimental apparatus consisted of five force transducers to measure tapping force by each finger. The subjects tapped force transducers for 7 sec as fast as possible using one (single-finger tapping) or two fingers (double-finger tapping). In the single-finger tapping the subjects were instructed to tap the designated force sensor by the index or ring finger, while in the double-finger tapping they use the index-middle finger- or ring-little finger-pair in an alternate manner. In both tasks, the subjects were instructed to keep "resting" (non-tapping) fingers in contact with their transducers. All subjects conducted twelve-day (three days/ week) training of all tapping tasks. In each of the training days, the subjects performed ten trials for each of the four tapping tasks. After each training session and a month after the final day of training, testing trials for data collection were performed for each task. In the elderly subjects, faster tapping was attained after the training with the most prominent effect on the ring-little finger-pair, followed by the index-middle finger-pair, the ring finger, and the index finger. The faster tapping compared with before training was also observed in all conditions a month after the final day of training. Thus, the training effects were maintained in spite of 1-month detraining. A significant change in tapping ability of the fingers was observed within two weeks of training. Therefore, observed improvement in tapping ability of the fingers is most likely due to changes in central neural function facilitating sequential motor action, which seems to be degenerated largely with aging. In addition, the changes might be maintained for a while without training.