Objective: There is increased need for identifying the early stages of AD to help facilitate emerging treatment. There is a question as to whether Free versus Cued Recall paradigms are more effective in early diagnosis of AD. The Loewenstein-Acevedo Scales of Semantic Interference and Learning (LASSI-L) is a novel list-learning test that assesses free versus cued recall in detecting subtle changes in cognition. Method: Eighty-six older adults (71% female) with a mean age of 75.5 (SD = 7.5 years) were studied. Twenty-nine participants with amnestic mild cognitive impairment (aMCI) and 57 cognitively normal (CN) controls were administered the LASSI-L. Participants were presented with List A (15 semantically-related words: fruits, musical instruments, and clothing), followed by free and cued recall trials of the same wordlist. Following presentation of List B (a content-equivalent wordlist), a second learning trial of List A was presented, and cued recall was assessed again. Results: Free and cued recall measures of maximum storage, and vulnerability to PSI and recovery from PSI, were entered into stepwise logistic regression models. Results indicated a combination of Cued B1 (p < .005) [vulnerable to PSI] and Cued A1 (p < .005) [tapping initial learning effects] were most predictive of separating aMCI from CN groups. Using area under ROC curve, sensitivity was 62%, and specificity yielded an overall correct classification rate of 82.6%. Conclusion: When using a controlled learning paradigm with semantic retrieval cues, cued recall was far superior to free recall in differentiating aMCI from CN participants. These findings have significant implications for future research.