Neuropsychological testing is important for the diagnosis and follow-up of dementia; it can also help provide consultation on patient care. However, lengthy testing is costly and can be stressful to the subjects. Tests developed for members of the majority culture often are inappropriate for ethnic minorities, especially those who speak a different language, have little or no formal education, and grew up in vastly different circumstances. Variables that directly affect test performance, such as education and acculturation instead of race or ethnicity, should be considered as explanatory variables for test performance. Reading level may be a better index for educational attainment than the number of years in school. Neuropsychological testing can be improved in several directions: (1) Use tests that are appropriate for the subject's background. (2) Use detailed scoring of a smaller number of test items to reduce test time, and establish discontinuation rules to limit the subject's experience of failure. (3) Develop parallel test forms for repeated assessment of the same individuals. (4) Strive for large and representative standardization samples. (5) Use computerized test norms based on findings of statistical analysis to better adjust for confounding variables.