A retrospective study including 515 Mycobacterium tuberculosis isolates from 215 patients was conducted to investigate possible laboratory contamination with M. tuberculosis over a 1-year period in a university hospital. All cultures underwent variable-number tandem-repeat (VNTR) typing. Cultures suspected of being contaminated in the VNTR analysis and possible sources of contamination underwent mycobacterial interspersed repetitive unit (MIRU) typing further. Overall, 8 (3.7%) cases of 215 patients were considered possible false-positives. Five (2.3%) cultures might be contaminated during initial batching processing, and 1 (0.5%) and 4 (1.9%) cultures of them were further classified as presumed and possible cases, respectively, of cross-contamination on clinical grounds. Three (1.4%) cultures might be contaminated by cultures that had been processed in species identification procedures in the same laminar-flow hood. The 2-step strategy using VNTR and MIRU analyses in combination in this study appears to be a valuable means for the study of false-positive cultures.