This research was designed to test the hypothesis that motor practice can enhance the capabilities of motor control in healthy controls (NC) and patients with a diagnosis of probable Alzheimer's disease (AD) and mild cognitive impairment (MCI), and consequently results in better motor performance. Approximately half of the subjects in the NC (n = 31), AD (n = 28), and MCI (n = 29) either received or did not receive practice on a task of fast and accurate arm movement with a digitizer. Changes in movement time (MT), movement smoothness (jerk), and percentage of primary submovement (PPS) were recorded and compared among the three groups across six blocks of trials (baseline and five training sessions). For all subjects, practice improved motor functions as reflected by faster and smoother motor execution, as well as a greater proportion of programming control. Compared to unaffected matched controls, AD and MCI subjects exhibited a greater reduction in movement jerk due to practice. Movement time and PPS data revealed that motor practice appeared to reduce the use of "on-line" correction adopted by the AD or MCI patients while performing the aiming movements. Evidently, their arm movements were quicker, smoother, and temporally more consistent than their untrained peers. The findings of this study shed light on how MCI and AD may affect motor control mechanisms, and suggest possible therapeutic interventions aimed at improving motor functioning in these impaired individuals.