Letter and category fluency tasks are used to assess semantic knowledge, retrieval ability, and executive functioning. They appear to be useful in detecting different types of dementia, but accurate detection of neuropsychological impairment relies on appropriate normative data. Multiple regression analysis was used to develop demographically corrected norms for letter and category fluency in 768 normal adults. T-score equations were developed on a base subsample of 403, and crossvalidated on a separate subsample (n = 365). Participants ranged in age from 20 years to 101 years; in educational level from 0 to 20 years; 55% were Caucasian and 45% were African American. Together, age, education, and ethnicity were significant predictors of letter and category fluency performance, accounting for 15% and 25% of variance, respectively. Formulas and tables for converting raw fluency scores to demographically corrected T scores are presented.