Mycobacteria cause a variety of illnesses that differ in severity and public health implications. The differentiation of Mycobacterium tuberculosis from nontuberculous mycobacteria (NTM) is of primary importance for infection control and choice of antimicrobial therapy. Despite advances in molecular diagnostics, the ability to rapidly diagnose M. tuberculosis infections by PCR is still inadequate, largely because of the possibility of false-negative reactions. We designed and validated a real-time PCR for mycobacteria by using the LightCycler system with 18 reference strains and 168 clinical mycobacterial isolates. All clinically significant mycobacteria were detected; the mean melting temperatures (with 99.9% confidence intervals [99.9% CI] in parentheses) for the different mycobacteria were as follows: M. tuberculosis, 64.35 degrees C (63.27 to 65.42 degrees C); M. kansasii, 59.20 degrees C (58.07 to 60.33 degrees C); M. avium, 57.82 degrees C (57.05 to 58.60 degrees C); M. intracellulare, 54.46 degrees C (53.69 to 55.23 degrees C); M. marinum, 58.91 degrees C (58.28 to 59.55 degrees C); rapidly growing mycobacteria, 53.09 degrees C (50.97 to 55.20 degrees C) or 43.19 degrees C (42.19 to 44.49 degrees C). This real-time PCR assay with melting curve analysis consistently accurately detected and differentiated M. tuberculosis from NTM. Detection of an NTM helps ensure that the negative result for M. tuberculosis is a true negative. The specific melting temperature also provides a suggestion of the identity of the NTM present, when the most commonly encountered mycobacterial species are considered. In a parallel comparison, both the LightCycler assay and the COBAS Amplicor M. tuberculosis assay correctly categorized 48 of 50 specimens that were proven by culture to contain M. tuberculosis, and the LightCycler assay correctly characterized 3 of 3 specimens that contained NTM.