The prevalence of tuberculous infection (i.e., reactions greater than or equal to 10 mm to 5 tuberculin units of purified protein derivative; Mantoux skin test) was determined among employees of 10 hospitals located throughout the United States. The risk of infection was strongly associated with age and race/ethnicity; nonwhites and older individuals were at higher risk. The prevalence of infection among hospitals varied threefold, from 7.0% to 21.4%. After adjusting for differences in the characteristics of employee groups (e.g., age, race/ethnicity, and sex), twofold differences among hospitals were still observed. The occurrence of "boosting" on retest was also studied. Among the different hospitals, the rate varied from 0% to nearly 10%. Race/ethnicity and age were the characteristics most closely associated with boosting. From our data and other data in the literature, we conclude that all hospitals should use two-step testing at least on a pilot basis. Our calculations suggest that two-step testing for employees over 35 years of age could be cost effective if the booster rate is greater than 1% of the employees retested.