We administered the Mattis Dementia Rating Scale (MDRS) to 1,001 healthy volunteers, aged 50 to 80 years, randomly selected from our community. Multivariate regression analysis revealed educational level (p = 0.000004) and age (p = 0.00001), but no other sociodemographic or risk factors for stroke, to be significantly associated with the MDRS score. The age- and education-specific lowest quintile cutoff scores ranged from 140 in subjects aged 50 to 59 years with at least college experience to 130 in subjects aged 70 to 80 years with only 4 to 9 years of schooling. These percentile distributions obtained for decades of age and different levels of education should be useful reference values for clinicians and investigators when applying the MDRS to assess cognitive functioning.