Objective: Dementia is a devastating neurological disease that may be better managed if diagnosed earlier when subclinical neurodegenerative changes are already present, including subtle cognitive decline and mild cognitive impairment. In this study, we used item-level performance on the Montreal Cognitive Assessment (MoCA) to identify individuals with subtle cognitive decline. Method: Individual MoCA item data from the Alzheimer's Disease Neuroimaging Initiative was grouped using k-modes cluster analysis. These clusters were validated and examined for association with convergent neuropsychological tests. The clusters were then compared and characterized using multinomial logistic regression. Results: A three-cluster solution had 77.3% precision, with Cluster 1 (high performing) displaying no deficits in performance, Cluster 2 (memory deficits) displaying lower memory performance, and Cluster 3 (compound deficits) displaying lower performance on memory and executive function. Age at MoCA (older in compound deficits), gender (more females in memory deficits), and marital status (fewer married in compound deficits) were significantly different among clusters. Age was not associated with increased odds of membership in the high-performing cluster compared to the others. Conclusions: We identified three clusters of individuals classified as cognitively unimpaired using cluster analysis. Individuals in the compound deficits cluster performed lower on the MoCA and were older and less often married than individuals in other clusters. Demographic analyses suggest that cluster identity was due to a combination of both cognitive and clinical factors. Identifying individuals at risk for future cognitive decline using the MoCA could help them receive earlier evidence-based interventions to slow further cognitive decline. Key Points Question: This study investigated a heterogenous group of individuals classified as cognitively unimpaired via cluster analysis of individual-item Montreal Cognitive Assessment (MoCA) performance. Findings: Individuals were organized into three clusters with 77.3% precision; the lowest performing cluster was characterized by lower scores on memory and executive function items, older age, and less common married status. Importance: This study sorted individuals categorized as cognitively unimpaired into three distinct clusters according to an easy-to-administer global cognitive test, thus providing a basis for clinicians to quickly and cost-effectively assess their patients' risk for future cognitive decline. Next Steps: Future research should focus on the neurodegenerative correlates of cluster membership, including amyloid-β deposition and cortical atrophy.