Objective: Early identification of COVID-19 in patients is important to prevent significant worsening of the disease. This study was undertaken to verify whether MEWS (Modified Early Warning Score), NEWS(National Early Warning Score), ROX index, and CURB-65, which are early diagnostic tools for severe respiratory diseases, could be applied to patients visiting the emergency room for COVID-19. Methods: This retrospective observational study included patients who visited an emergency medical center from September 1 to October 31, 2020, and from January 1 to February 28, 2021. Based on the vital signs and blood tests during the emergency room visit, severity evaluation tools and early diagnostic tools for severe cases were used and compared according to their area under the curve (AUC) values. The primary outcome was in-hospital mortality, while the secondary outcomes were intensive care unit admission rate and the need for mechanical ventilation based on these four tools (MEWS, NEWS, ROX index, and CURB-65). Results: A total of 667 patients were analyzed. No significant difference was determined between the non-survivor group and survivor group in the MEWS values (P=0.13), but statistically significant differences were observed for NEWS (5 vs. 1, P<0.05), CURB-65 (2 vs. 1, P<0.05), and ROX index (16.61 vs. 23.1, P<0.01). The AUC value of NEWS for death prediction indicated a good predictive power at 0.80, while that of MEWS showed a low predictive power at 0.57, which was statistically significant. Moreover, the AUC values of CURB-65 and ROX index did not differ significantly from values obtained for NEWS. Conclusion As early diagnostic tools for predicting death in COVID-19 patients, NEWS, ROX index, and CURB-65 showed excellent discrimination ability, whereas MEWS showed statistically and significantly lower discrimination ability.