Normative data for the Mattis Dementia Rating Scale (MDRS), stratified by age and education, are provided for use with older adults (ages 61-94) in urban medical settings. Age and education accounted for the greatest amount of variance in MDRS performance. Gender and race were also associated with total MDRS scores, though to a lesser extent. The present normative data are more appropriate for use with older patients seen in urban medical settings than normative data obtained from samples of better-educated, relatively healthy, Caucasian adults. This study provides additional evidence of the significant influence of age and education on MDRS total score (MDRS-T) performance, and highlights the importance of matching an examinee's demographic background to the normative sample with which his or her test score is being compared. Careful consideration of this match is likely to lead to more accurate diagnostic conclusions and potentially improved patient care.