In the present study, we aimed to identify biomarkers for diagnosis of pulmonary tuberculosis (PTB) using urinary metabolomic and proteomic analysis. Methods: 40 urine samples were collected from PTB, lung cancer (LCA), community-acquired pneumonia (CAP) patients and healthy controls (HC), respectively. Biomarker panels were selected based on random forest (RF) analysis. Results: A total of 3,868 proteins and 47,528 metabolic features detected using pairwise comparisons. Using AUC≥0.8000 as a cutoff value, we picked up five protein biomarkers for PTB diagnosis. The five-protein panel yielded an AUC of PTB/HC, PTB/CAP and PTB/LCA were 0.9840, 0.9680 and 0.9310, respectively. Additionally, five metabolism biomarkers were selected for differential diagnosis purpose. By employment of the five-metabolism panel, we could differentiate PTB/HC at an AUC of 0.9940, PTB/CAP of 0.8920, and PTB/LCA of 0.8570, respectively. When the five protein and five metabolism biomarkers were combined, yielded an AUC of PTB/HC, PTB/CAP and PTB/LCA were 1.0000, 0.9220 and 0.9500, respectively. Conclusion: Our data demonstrate that the combination of metabolomic and proteomic analyses can identify a novel urine biomarker panel to diagnose PTB with high sensitivity and specificity.