Objective: The current study investigated differential practice effects within a single testing session in older individuals with Mild Cognitive Impairment (MCI), Alzheimer's disease (AD), and healthy controls. Method: Participants included 90 patients with MCI, 23 with AD, and 42 cognitively intact controls. All were administered the Hopkins Verbal Learning Test – Revised Total Recall (HVLT), semantic fluency, and letter fluency tasks twice within the same evaluation (approximately 60 minutes apart). Results: Controls significantly improved on semantic fluency and HVLT across this brief retest interval. The MCI participants also improved, but to a lesser degree, and the AD patients did not improve on any measures. Using the control data, standardized regression-based change formulae were developed. Not surprisingly, controls were largely stable across measures, whereas more than one third of MCI and AD patients evidenced reliable decline on the HVLT. Conclusion(s): Differential practice effects within the same testing session were found between the groups, with controls showing the greatest improvement on retesting and AD patients showing the least. The regression-based change models highlighted these differences, and they could prove to be useful in clinical and research settings to determine the magnitude of practice in individual patients/subjects. Practice effects, elicited within a single testing session and with relatively little additional clinical effort, may provide valuable information about diagnosis, prognosis, and treatment recommendations in memory-disordered patients. Additional studies might vary the retest interval to determine the length of time most sensitive to practice effects.