Tuberculosis (TB) remains a major global health issue, especially in resource-limited settings, where rapid and accurate diagnosis is critical yet often lacking. Nuclear magnetic resonance (NMR)-based metabolomics has emerged as a promising tool for identifying disease-specific biomarkers. This study assessed the diagnostic potential of urinary metabolomics using high-resolution (HR-)NMR and benchtop NMR (bNMR) in Nigerian adults with presumed pulmonary TB, including individuals with and without HIV. Urine samples were analysed by HR-NMR and bNMR. Multivariate analysis was used to develop classification models based on confirmed TB status. Key discriminant metabolites were identified. In HIV-negative individuals, HR-NMR achieved 84% sensitivity and 86% specificity, while bNMR reached 83% and 84%, respectively. In people living with HIV, HR-NMR reached 75% sensitivity and 70% specificity; bNMR achieved 90% and 46%. Both models met or exceeded WHO sensitivity criteria for non-sputum-based TB diagnostics. Thirteen discriminatory metabolites were identified, with eight consistent with previous studies. These findings suggest common metabolic signatures of TB, regardless of HIV status. Urinary NMR-based metabolomics shows strong potential for TB diagnosis in high TB/HIV burden settings. HR-NMR provides excellent accuracy, while bNMR offers a cost-effective alternative. Further studies are needed to validate and optimise this approach for clinical use.