The Gen-Probe amplified Mycobacterium tuberculosis direct test can give discrepant results directly in respiratory or cultured samples from patients infected with Mycobacterium celatum, leading to inappropriate therapy for, in our case, an immunocompetent patient.Mycobacterium celatum was first described in 1993 (3), and since then, sporadic reports have been published on the isolation of this mycobacterium from immunocompromised patients (1,5,6,8,11,12 1994), and a fatal pulmonary infection in an apparently healthy adult (4). Biochemically, the organism is indistinguishable from the Mycobacterium avium complex, and mycolic acid high-pressure liquid chromatography analysis or genetic analysis is required for proper identification (3). For identification, cross-reactivity with nontuberculous mycobacteria has been described with the AccuProbe culture confirmation test (Gen-Probe, San Diego, Calif.) for the M. tuberculosis complex (2, 10). This note describes the initial misidentification and interpretation of a positive amplified M. tuberculosis direct (AMTD) test (GenProbe) result for a respiratory sample derived from an immunocompetent patient with an M. celatum infection.A 61-year-old man was referred on 5 March 1999 by a general practitioner to the outpatient respiratory disease clinic because of general malaise, a productive cough, and an unexplained 16-kg weight loss in the past 3 years. The chest radiograph showed a pulmonary infiltrate in the right upper lobe, with extensive cavities identified by computed tomography. A bronchial lavage was performed and showed on average more than one acid-fast rod per high-power field (magnification, ϫ1,000. The AMTD test was positive with a value of 1,000,000 relative light units (RLU). Tuberculostatic drug therapy was started with 300 mg of isoniazid, 600 mg of rifampin, and 2,000 mg of pyrazinamide. Liquid culture (MB/BacT; Organon Teknika, Boxtel, The Netherlands) was positive on 25 March 1999 and was subcultured on Löwenstein Jensen agar. The subculture grew very small smooth colonies with a pale yellow pigment and was sent to the Mycobacterium section of the National Institute of Public Health and the Environment (RIVM, Bilthoven, The Netherlands) for drug susceptibility testing by the agar proportion method and species identification. The AccuProbe culture confirmation tests for M. tuberculosis, M. avium complex, Mycobacterium kansasii, and Mycobacterium gordonae were negative. Because of the discrepancy between the AMTD test as performed by the Medical Microbiology department at the University Hospital Maastricht and the AccuProbe results of the RIVM, a culture was sent again to the RIVM. This material again yielded a positive result (509,094 RLU) in the AMTD test in Maastricht and again yielded a negative result in the AccuProbe test at the RIVM. Therefore, it was concluded that the AMTD result was probably a false positive. Drug susceptibility data were only obtained on 26 June 1999 (Table 1) because of the slow growth and the initial suspicion that we were de...