1. Adult owl monkeys were trained to detect a difference in the frequency of sequentially applied tactile stimuli presented to a constant, restricted location on the glabrous skin of a single finger. Psychophysical performance functions and thresholds were determined on daily sessions over a 3- to 20-wk-long training period. 2. Thresholds for the trained digit progressively decreased from a 6- to 8-Hz difference to a 2- to 3-Hz difference relative to a 20-Hz standard. These thresholds were similar to those described for macaques and humans determined by the use of a two-alternative forced-choice procedure. 3. Six of the seven studied monkeys showed a continuously progressive improvement in performance with training. Early in the training period, the performance improved at about the same rate for all frequencies. Later in the training period, the performance for frequencies much greater than the comparison frequency improved sooner than did the performances for frequencies more similar to the comparison frequency. This resulted in an increase of the slope of the psychometric function near threshold. In a single monkey, no clear later-component improvements were recorded. 4. Analyses of performances using the theory of signal detection revealed a progressive increase in the measure of d' for all frequencies above threshold. 5. Some improvements in performance were also recorded when stimuli were applied on an adjacent digit, which was trained for 2 or 3 sessions spaced throughout the course of training. However, thresholds on these digits were always greater than those on the trained digit. These findings suggest that there are local changes generated by this training at somatotopically restricted regions of the central somatosensory nervous system. 6. It is concluded that this training resulted in a genuine progressive improvement in temporal acuity specific to the trained skin. The initial rapid improvement was likely due to an improvement in the "strategy" or "cognitive" aspects of the task, whereas more gradual improvements in performance recorded throughout the training period were most probably due to somatotopically localized changes in the neural representations of the behaviorally relevant stimulus.