OBJECTIVES: To document the actual tuberculosis (TB) control policies and procedures in a nonoutbreak setting in a variety of hospitals. To determine if any particular practices are linked to higher rates of employee tuberculin skin-test conversion. DESIGN: Survey of hospital occupational health and infection control practitioners for the year 1994 regarding hospital TB policies. Review of hospital records to verify the number of patients with TB at each hospital and to verify the number of employees with positive tuberculin skin tests. Smoke-stick testing of negative-pressure ventilation rooms. SETTING: A 13-hospital health system in the Midwest. RESULTS: Hospitals ranged in size from 40 to 1,208 beds (median 220) and employed 150 to 6,500 workers (median 875). There were seven rural and six urban centers, including four teaching hospitals. All 13 hospitals had TB control plans, and all performed annual tuberculin skin testing on employees. Annual skin-test positivity rates ranged from 0% to 1.0% (median 0.3%). Negative-pressure ventilation rooms were available in 11 hospitals. The percentage of negative-pressure rooms with effective negative pressure ranged from 44% to 100% (median 95%). Three of the 13 hospitals used high-efficiency particulate air (HEPA) masks as primary personal respiratory protection, and 8 used dust-mist or dust-mist-fume masks. We found no relation between the type of face mask used, number of functional negative-pressure rooms, or hospital TB risk category, and employee skin-test conversion rates. CONCLUSIONS: Considerable variation existed in the TB control policies and procedures between hospitals, but employee TB skin-test conversion rates were low in all settings.