Between Jan 1, 1981, and Aug 31, 1983, 1,323 Ohio State University Hospital employees were screened tor hepatitis B. There were 105 employees (cases) with present or past evidence of hepatitis B virus infection, and 210 randomly selected employees (controls) with no evidence of hepatitis B virus infection. Using logistic regression analyses to develop a risk index, the best predictive model indicated that nonwhites, males, discontinuity of employment, and frequency of contact with blood products constituted the most important risk factors in acquiring hepatitis B virus infections. Those with no blood product contact were at higher risk than those exposed to blood products. Thus, although there was a gradient of risk among those exposed to blood products, the findings suggest other factors may contribute more substantially to the risk than the exposure variables tested.