Paired associate learning tasks are reportedly particularly sensitive to preclinical Alzheimer's disease. We aimed to determine the effectiveness of the recently updated Wechsler Memory Scale verbal paired associates (VPA) in distinguishing the earliest stages of memory impairment (amnestic mild cognitive impairment, aMCI), and the clinical application at the case level, compared with other episodic memory tasks. Participants were 77 people with aMCI and 77 matched healthy older adults (HOA). VPA performance distinguished aMCI from HOA at the group level with large effect sizes, of similar size to the other tasks at immediate recall, but smaller than the CVLT-II list-learning task at delayed recall. Similarly, receiver operating characteristic (ROC) analysis demonstrated good discrimination, similar to other tasks, but again with CVLT-II more accurate at delayed recall. Although group differences remained for normative data, on a case basis using existing normative data the VPA failed to identify 70% of aMCI as impaired. The findings suggest further examination of the normative data is required before the VPA is useful in clinical practice, and highlight the importance of comprehensive neuropsychological assessment in detecting mild memory changes in older adults.