The Trail Making Test (TMT) is a useful measure of executive dysfunction in elderly subjects. This study aims to investigate the discriminative validity of the oral version of the TMT (OTMT), which can be administered to subjects with visual or motor disabilities, in elderly patients with Mild Cognitive Impairment (MCI; n = 30), Alzheimer's disease (AD; n = 30), and healthy controls (HCs; n = 25). The WAIS-R Digit Span Backwards Subscale, written form of the Trail Making Task, the Clock Drawing Test, the AD Assessment Scale-Cognitive Subscale, and the OTMT were also administered to all participants in order to examine the concurrent validity of the OTMT. The OTMT part B discriminated between patients with MCI, AD, and HC correctly. The OTMT completion time was not correlated with age, but was negatively correlated with education. In conclusion, the OTMT (mostly part B) is a valid and practical measurement tool for different levels of cognitive impairment, especially for patients with visual or motor disabilities for whom the classical written form is not feasible.