Background: The number and significance of tuberculin skin test reactions were compared with self-reported baseline values among house staff working in a public hospital. High-risk medical specialties, locations, and infection control practices were examined. Methods: House start interviews, tuberculin skin test applications, review of employee health service records, and environmental monitoring of high-risk areas were performed. Results: Among house staff self-reported as having negative tuberculin skin test status, 46.2% (95% CI 27.0% to 65.4%) of internal medicine house staff, compared with 4.8% (95% CI 4.3% to 13.9%) of house staff from other areas (p < 0.005), had positive results on a repeat tuberculin skin testing before graduation. These differences were not entirely explained by the use of surgical masks, year of training, or previous vaccination with bacille Calmette-Guerin. Most skin test reactions (69%) occurred among house staff who had not been vaccinated with bacille Calmette-Guerin. Increased skin reactivity probably represented excess conversions from unprotected exposure. Tuberculosis transmission was facilitated by delays in diagnosis, inadequate isolation facilities, and suboptimal ventilation. House staff did not comply with recommended tuberculosis surveillance because of time constraints, fear, and misunderstandings about tuberculin skin test interpretations in light of previous bacille Calmette-Guerin vaccination. Conclusions: House staff in high-exposure settings with suboptimal environmental controls are at increased risk for tuberculosis infection. Participation in surveillance programs can be increased by enlisting the participation and advocacy of respected medical colleagues, screening house staff differentially according to exposure and job classifications, and more accurately interpreting subsequent test results from baseline two-step testing.