BACKGROUND: Variability of finger tapping speeds, especially in the non-dominant hand, has been reported in individuals with amnestic mild cognitive impairment (MCI-A) and dementia of the Alzheimer's type (AD). An explanation of this finding, however, has not appeared. OBJECTIVE: The aim of this study was to investigate possible motor correlates of finger tapping variability in normal older healthy controls (HC), persons with subjective memory complaints (SMC), MCI-A and probable AD. METHODS: Using a modified version of the Halstead Finger Tapping Test (HFTT), individual finger tapping movements were classified as "valid" (i.e., advanced the number on a mechanical counter) or "invalid" (i.e., did not advance the number). Failures at selective motor inhibition and learning and fatigue effects were also measured. RESULTS: Finger tapping variability was significantly greater in the non-dominant hand in probable AD patients compared to HC and SMC patients. MCI-A and probable AD patients did not differ on this measure. Finger tapping variability was significantly correlated (ρ = + 0.65) with the number of invalid tapping responses but not with selective motor inhibition failures. A small but significant correlation of dominant hand learning effect with finger tapping variability was found. Invalid tapping responses were significantly greater in MCI and probable AD groups compared to HC and SMC groups. CONCLUSIONS: Frequency of invalid tapping responses strongly correlated with finger tapping variability. The number of invalid tapping responses may have diagnostic and prognostic significance when evaluating older individuals with known or suspected memory impairment of a neurodegenerative type.