The Montreal Cognitive Assessment (MoCA) was recently proposed as a cognitive screening test for milder forms of cognitive impairment, having surpassed the well-known limitations of the Mini-Mental State Examination (MMSE). This study aims to validate the MoCA for screening Mild Cognitive Impairment (MCI) and Alzheimer disease (AD) through an analysis of diagnostic accuracy and the proposal of cut-offs. Patients were classified into 2 clinical groups according to standard criteria: MCI (n=90) and AD (n=90). The 2 control groups (C-MCI: n=90; C-AD: n=90) consisted of cognitively healthy community dwellers selected to match patients in sex, age, and education. The MoCA showed consistently superior psychometric properties compared with the MMSE, and higher diagnostic accuracy to discriminate between MCI (area under the curve=0.856; 95% confidence interval, 0.796-0.904) and AD patients (area under the curve=0.980; 95% confidence interval, 0.947-0.995). At an optimal cut-off of below 22 for MCI and below 17 for AD, the MoCA achieved significantly superior values in comparison with MMSE for sensitivity, specificity, positive predictive value, negative predictive value, and classification accuracy. Furthermore, the MoCA revealed higher sensitivity to cognitive decline in longitudinal monitoring. This study provides robust evidence that the MoCA is a better cognitive tool than the widely used MMSE for the screening and monitoring of MCI and AD in clinical settings.