To characterize the prevalence of geographic tongue (GT) among US adults. Population-based case-control study. The presence or absence of GT. Data from 16 833 adults examined during The Third National Health and Nutrition Examination Survey, 1988-1994 (NHANES III), a study based on multistage probability sampling were analyzed using SAS-callable SUDAAN 9.0.1. Geographic tongue point prevalence was 1.8% (95% CI: 1.4, 2.3). Multivariate logistic regression showed significant effects of race-ethnicity, with Whites (AOR = 1.8; 1.3, 2.5) and Blacks (AOR = 1.6; 1.2, 2.1) having greater odds of GT than Mexican-Americans; current corticosteroid therapy (AOR = 3.7; 1.54, 8.6). Cigarette smokers had lower GT prevalence (AOR = 0.4; 0.3, 0.6). Fissured tongue (FT) was strongly associated with GT among non-smokers: AOR = 17.5 (7.8, 39.5). We did not find significant associations with age, gender, oral contraceptive use, diabetes mellitus, allergy or atopy, psychological or dermatological conditions as previous research has suggested. Geographic tongue was more prevalent among Whites and Blacks compared with Mexican-Americans, positively associated with FT, and inversely associated with cigarette smoking.