Abstract: While finger tapping speed, using the traditional Halstead Finger Tapping Test (HFTT), has not separated normal age matched controls with those with mild cognitive impairment (MCI) or early Alzheimer’s Dementia (AD), variability of finger tapping speeds may differ between these groups. Using a modified version of the HFTT, we demonstrated that finger tapping variability was significantly greater in the non-dominant hand in AD patients compared to health controls (HC) and patients with subjective memory complaints (SMC). We also found that finger tapping variability was highly correlated (rho= +0.65) with the number of “invalid” tapping responses (i.e., index finger movements did not advance the number on a mechanical counter). The total number of “invalid” tapping responses in the dominant and non-dominant hands significantly separated SMC vs MCI vs AD. There were no significant differences between HC and SMC groups. The number of “invalid” tapping responses negatively correlated with overall neuropsychological status as measured by the BNI Screen for Higher Cerebral Functions (rho=-0.51). The total number of “invalid” responses was not associated with the educational level. Therefore, the number of “invalid” tapping responses may be a useful diagnostic tool when evaluating older individuals with memory complaints irrespective of their educational level.