In spite of the decrease in the number of registered leprosy patients, the number of new cases diagnosed each year (400,000) has remained essentially unchanged. Leprosy diagnosis is difficult due to the low sensitivity of current methodologies to identify new cases. In this study, conventional and TaqMan real-time PCR assays for detection of Mycobacterium leprae DNA were compared to current classification based on clinical, bacteriological, and histological evaluation. M. leprae DNA was extracted from frozen skin biopsy specimens from 69 leprosy patients enrolled in the study and was amplified using specific primers for either the antigen 85B-coding gene or the 85A-C intergenic region by using conventional and real-time PCR. The detection rate was 100% among multibacillary (MB) patients and ranged from 62.5% to 79.2% among paucibacillary (PB) patients according to the assay used. The TaqMan system for 85B gene amplification showed the highest sensitivity, although conventional PCR using the 85A-C gene as a target was also efficient. The cycle threshold (C(T)) values obtained using the TaqMan system were able to statistically (P < 0.0001) differentiate MB (mean C(T), 28.06; standard deviation [SD], 4.51) from PB (mean C(T), 33.06; SD, 2.24) patients. Also, there was a correlation between C(T) values and the bacteriological index for MB patients (Pearson's r, -0.444; P = 0.008). Within the PB patients' group, we tested normal skin from six patients exhibiting the pure neuritic form of leprosy (PNL). Five out of six PNL patients were positive for the presence of M. leprae DNA, even in the absence of skin lesions. In conclusion, the TaqMan real-time PCR developed here seems to be a useful tool for rapidly detecting and quantifying M. leprae DNA in clinical specimens in which bacilli were undetectable by conventional histological staining.