Bacille Calmette-Guérin (BCG) vaccination can confound tuberculin skin test (TST) reactions in the diagnosis of latent tuberculosis infection (LTBI). The TST was compared with a Mycobacterium tuberculosis (MTB)-specific enzyme-linked immunospot (ELISPOT) assay during an outbreak of MTB infection at a police academy in Germany. Participants were grouped according to their risk of LTBI in close (n = 36) or occasional (n = 333) contacts to the index case. For the TST, the positive response rate was 53% (19 out of 36) among close and 16% (52 out of 333) among occasional contacts. In total, 56 TST-positive contacts (56 out of 71 = 78.9%) and 27 TST-negative controls (27 out of 298 = 9.1%) underwent ELISPOT testing. The odds ratio (OR) of a positive test result across the two groups was 29.2 (95% confidence interval (CI) 3.5-245.0) for the ELISPOT and 19.7 (95% CI 2.0-190.2) for the TST with a 5 mm cut-off. Of 369 contacts, 158 (42.8%) had previously received BCG vaccination. The overall agreement between the TST and the ELISPOT was low, and positive TST reactions were confounded by BCG vaccination (OR 4.8 (95% CI 1.3-18.0)). In contrast, use of a 10-mm induration cut-off for the TST among occasional contacts showed strong agreement between TST and ELISPOT in nonvaccinated persons. In bacille Calmette-Guérin-vaccinated individuals, the Mycobacterium tuberculosis-specific enzyme-linked immunospot assay is a better indicator for the risk of latent tuberculosis infection than the tuberculin skin test.