Metagenomic Next-Generation Sequencing (mNGS) holds promise in diagnosing fever of unknown origin (FUO) by detecting diverse pathogens. We systematically reviewed the literature to evaluate mNGS's accuracy, clinical efficacy, and limitations in FUO diagnosis. Nine studies revealed mNGS's positivity rate ranging from 66.7% to 93.5% for bacterial bloodstream infections and systemic infections. Meta-analysis of three studies involving 857 patients, including 354 with FUO, showed a sensitivity of 0.91 (95% CI: 0.87-0.93) and specificity of 0.64 (95% CI: 0.58-0.70). Despite lower specificity, mNGS demonstrated a higher Diagnostic Odds Ratio (DOR) of 17.0 (95% CI: 4.5-63.4) compared to conventional microbiological tests (CMTs) at 4.7 (95% CI: 2.9-7.6). While mNGS offers high sensitivity but low specificity in identifying causative pathogens for FUO, its superior DOR suggests potential for more accurate diagnoses and targeted interventions. Further research is warranted to optimize its clinical application in FUO management.