Objective: First, we expected to replicate reports that older adults are impaired in reversal learning abilities compared to younger adults. Next, we aimed to determine if reversal learning abilities differed when cognitively normal older adults are dichotomized into impaired and unimpaired subsets. Method: Participants were 32 younger adults (Age: M = 20.88, SD = 2.76; Men = 8) and 29 older adults (Age: M = 72.62, SD = 8.11; Men = 2) who completed a battery of standardized neuropsychological tests including the Hopkins Verbal Learning Test-Revised (HVLT-R). Then, all completed the Reversal Task (RT) in which contingencies were learned and then reversed. Older adults were classified as impaired or unimpaired based on performance on the delayed-recall subtest of the HVLT-R. Results: Older adults committed more errors (Cohen's d = 0.99; p < .001) and required more trials to complete (d = 0.77; p = .004) the RT than younger adults. Importantly, impaired older adults committed more errors (d = 0.94; p < .001) and required more trials to complete (d = 1.14; p < .001) the RT than unimpaired older adults and younger adults. Younger adults and unimpaired older adults did not differ in RT performance (p > .05). Conclusion: We replicated that older adults were deficient in reversal learning abilities compared to younger adults. Further, we provided evidence that reversal learning abilities varied in older adults; specifically, only impaired older adults were deficient in reversal learning abilities. Forthcoming work should determine if deficits in impaired older adults are related to pathological cognitive aging.