The ability of diagnose the tuberculosis infection is an essential factor in the spreading control of tuberculosis. However, microscopic examination presents a low sensitivity and culture techinques require incubation times up to two months. This study aimed at developing a NMR-based metabolomic approach for the differential diagnosis of tuberculosis in urine samples. We examined samples from patients diagnosed of tuberculosis (n=19), other respiratory infection (n=25) and healthy controls (n=29). Unsupervised PCA provide a nearly perfect discrimination between the three groups. We identified 31 chemical shifts regions to develop predictive models for the diagnostic of tuberculosis, obteining an accuracy of 100%.