Verbal fluency tests are commonly used in neurocognitive and mental status examinations in patients with suspected dementia. Inflation of test scores as a result of practice effects may yield false-negative results in test-retest and multidisciplinary settings, particularly among patients with mild cognitive deficits. To address this issue, animal naming was administered twice within a 1-week period to a group of individuals referred for suspected dementia who were ultimately diagnosed with mild cognitive impairment (MCI; amnestic form), probable Alzheimer disease (AD), or no dementia. A 2 x 3 repeated-measures analysis of variance revealed a statistically significant interaction between administration time and group. Post hoc analyses indicated that nondemented controls were the only group to demonstrate a significant practice effect, producing an average of approximately three more animal names at time two. Like patients with a diagnosis of AD, subjects with amnestic MCI failed to benefit from repeated exposure to the animal naming test, and only controls showed an average improvement upon retest. This underscores the cognitive similarity between individuals diagnosed with amnestic MCI and AD and suggests that improvement upon retest may be a diagnostically useful finding.