Our objective was to evaluate the utility of the Montreal Cognitive Assessment (MoCA) in both non-clinical and clinical populations. The MoCA, a tool used widely in clinical geriatric practice to detect and quantify cognitive impairment, was administered to three diverse samples to assess the psychometric properties of the instrument. Participants were 482 healthy persons seen in the Cooper Center Longitudinal Study (CCLS), 1923 healthy volunteers evaluated by the Dallas Heart Study (DHS), a population-based sample of Dallas County residents and 69 persons with known or suspected brain pathology, seen for clinical evaluation. The standard deviations in the CCLS and DHS groups were small (2.81, 3.87) and moderate in the clinical group (5.22). The corresponding raw coefficients alpha were. 50,. 63, and. 75. Thus, variability and internal consistency were associated. MoCA scores generally did not correlate with gender and correlated with age only in the clinical group. However, there were moderately strong positive correlations between MoCA scores and education in all three samples. The reliability of the MoCA was extremely low in the two non-clinical groups, but was high in the clinical group. Although useful in large non-clinical studies, the MoCA is best suited to detect and quantify cognitive impairment in clinical patients.