Precise estimates of the incidence of hepatitis B virus (HBV) infection are required to assess the impact of immunization and other prevention strategies in the United States. Race- and age-specific prevalence data obtained from the second and third National Health and Nutrition Examination Surveys (NHANES II, 1976-1980, and NHANES III, 1988-1994) were used to estimate the annual incidence of HBV infection by catalytic modeling. During the period covered by NHANES II, an estimated 323,462 persons were infected annually, and 334,863 were infected annually during the period covered by NHANES III. No statistically significant declines in prevalence of HBV infection occurred between the two surveys, a period during which hepatitis B vaccination targeted only limited numbers of high-risk adults.