BACKGROUND: Among African-Americans, and in southern US states, the rates of stroke are high but the reported prevalences of atrial fibrillation (AF) are low. We hypothesized that the reported ethnic and regional distributions of AF are affected by the sensitivity of the methods that were used to detect AF in previous reports. METHODS: A total of 18,833 black and white participants from the US national REasons For Geographic And Racial Differences In Stroke (REGARDS) study were included in this analysis. Levels of sensitivity to detect AF, from least to most sensitive, were created for combinations of self-report (SR) and ECG methods, as follows: (1) SR plus ECG, (2) ECG alone, (3) SR alone, and (4) SR or ECG. Geographic regions were dichotomized as Stroke Belt (the southern US states) and non-Stroke Belt. Logistic regression analysis estimated the odd ratios of AF associated with the Stroke Belt and black ethnicity for each diagnostic combination. RESULTS: Residence in the Stroke Belt was significantly associated with AF when diagnosed by SR plus ECG (multivariable-adjusted OR, 0.66; 95% CI, 0.47 to 0.92), but not when diagnosed with SR or ECG (OR, 0.95; 95% CI, 0.85 to 1.06). Similarly, for the 4 methods used to detect AF, the strength of the association between black ethnicity and AF progressively decreased with increasing test sensitivity (ORs: 0.20, 0.40, 0.70, 0.71, respectively). CONCLUSIONS: The association of AF with residence in the Stroke Belt and black ethnicity was inversely related to the sensitivity of the method used to detect AF: as test sensitivity increased, the association became attenuated. This may partially explain the lower reported prevalence of AF in populations and regions with higher stroke rates.