Objective: Early identification of patients at risk for deterioration is crucial to reduce in-hospital mortality. Various early warning scores have been widely applied in the emergency department (ED) of hospitals. This study evaluates and compares the effectiveness of three early warning scores_Modified Early Warning Score, Rapid Acute Physiology Score (RAPS), Worthing Physiological Scoring System (WPS), and Rapid Emergency Medicine Score. These scores help predict the need for critical care and 24- and 72-hour mortalities among alert patients presenting to the ED with dyspnea. Methods: This retrospective cohort study used data from electronic medical records of patients admitted between 2018 and 2020 and included all consecutive alert patients who presented with dyspnea in the ED. The primary outcome was to evaluate the performance of early warning scores regarding the need for critical care. The secondary outcomes were the prediction of 24- and 72-hour in-hospital mortalities. Results: Among 4,322 patients evaluated, 255 received critical care, and 17 and 84 died within 24 and 72 hours, respectively. The WPS had the overall highest performance for predicting the need for critical care (area under the curve [AUC], 0.781; 95% confidence interval [CI], 0.751-0.810) and 24-hour (AUC, 0.816; 95% CI, 0.738-0.894) and 72-hour mortalities (AUC, 0.794; 95% CI, 0.750-0.838), but differed significantly only from the RAPS. Conclusion The WPS might better predict the need for critical care and short-term mortality in alert patients with dyspnea in the ED. However, owing to a lack of its superiority in statistics, further studies are warranted to conclude the optimal tools applicable for these patients.