The uniform multidrug therapy clinical trial, Brazil (U-MDT/CT-BR), database was used to describe and report the performance of available tools to classify 830 leprosy patients as paucibacillary (PB) and multibacillary (MB) at baseline. In a modified Ridley and Jopling (R&J) classification, considering clinical features, histopathological results of skin biopsies and the slit-skin smear bacterial load results were used as the gold standard method for classification. Anti-phenolic glycolipid-I (PGL-I) serology by ML Flow test, the slit skin smear bacterial load, and the number of skin lesions were evaluated. Considering the R&J classification system as gold standard, ML Flow tests correctly allocated 70% patients in the PB group and 87% in the MB group. The classification based on counting the number of skin lesions correctly allocated 46% PB patients and 99% MB leprosy cases. Slit skin smears properly classified 91% and 97% of PB and MB patients, respectively. Based on U-MDT/CT-BR results, classification of leprosy patients for treatment purposes is unnecessary because it does not impact clinical and laboratories outcomes. In this context, the identification of new biomarkers to detect patients at a higher risk to develop leprosy reactions or relapse remains an important research challenge.