A steadiness-improving intervention was used to determine the contribution of variability in motor unit discharge rate to the fluctuations in index finger acceleration and manual dexterity in older adults. Ten healthy and sedentary old adults (age 72.9 +/- 5.8 yr; 5 men) participated in the study involving abduction of the left index finger. Single motor unit activity was recorded in the first dorsal interosseus muscle before, after 2 wk of light-load training (10% maximal load), and after 4 wk of heavy-load training (70% maximal load). As expected, the light-load training was effective in reducing the fluctuations in index finger acceleration during slow shortening (0.25 +/- 0.12 to 0.13 +/- 0.08 m/s(2)) and lengthening contractions (0.29 +/- 0.10 to 0.14 +/- 0.06 m/s(2)). Along with the decline in the magnitude of the fluctuations, there was a parallel decrease in the coefficient of variation for discharge rate during both contraction types (33.8 +/- 6.8 to 25.0 +/- 5.9%). The heavy-load training did not further improve either the fluctuations in acceleration or discharge rate variability. Furthermore, the manual dexterity of the left hand improved significantly with training (Purdue pegboard test: 11 +/- 3 to 14 +/- 1 pegs). Bivariate correlations indicated that the reduction in fluctuations in motor output during shortening (r(2) = 0.24) and lengthening (r(2) = 0.14) contractions and improvement in manual dexterity (r(2) = 0.26) was directly associated with a decline in motor unit discharge rate variability. There was a strong association between the fluctuations in motor output and manual dexterity (r(2) = 0.56). These results indicate that practice of a simple finger task was accompanied by a reduction in the discharge rate variability of motor units, a decrease in the fluctuations in motor output of a hand muscle, and an improvement in the manual dexterity of older adults.