Objective To investigate whether a simplified inflammation-based risk scoring system comprising three readily available biomarkers (albumin, C-reactive protein, and leukocytes) may predict major adverse outcomes in patients with COVID-19. Methods Upon admission to the emergency room, the inflammation-based risk scoring system was applied and patients were classified as having mild, moderate, or severe inflammation. In-hospital occurrence of thrombosis, need for mechanical ventilation, and death were recorded. Results One-hundred patients (55 ± 13 years; 71% men) were included and classified as having mild (29%), moderate (12%), or severe (59%) inflammation. The need for mechanical ventilation differed among patients in each group (16%, 50%, and 71%, respectively; P < 0.0001), yielding a 4.1-fold increased risk of requiring mechanical ventilation in patients with moderate inflammation and 5.4 for those with severe inflammation. On the contrary, there were no differences for the occurrence of thrombosis (10%, 8%, and 22%, respectively; P = 0.142) or death (21%, 42%, and 39%, respectively; P = 0.106). In the multivariate analysis, only severe inflammation (hazard ratio [HR] = 4.1), D-dimer > 574 ng/mL (HR = 3.0), and troponin I ≥ 6.7 ng/mL (HR = 2.4) at hospital admission were independent predictors of the need for mechanical ventilation. Conclusion The inflammation-based risk scoring system predicts the need for mechanical ventilation in patients with severe COVID-19.