To compare the effectiveness of thromboembolic risk scores in determining in-hospital events of COVID-19 patients. This retrospective study included a total of 410 consecutive COVID-19 patients. Scores including CHADS-VASc-HS (congestive heart failure, hypertension, age, diabetes mellitus, stroke/transient ischemic attack, vascular disease, sex, hyperlipidemia, smoking); modified RCHADS-VASc (CHADS-VASc plus renal function), m-ATRIA (modified Anticoagulation and Risk Factors in Atrial Fibrillation score), ATRIA-HSV (ATRIA plus hyperlipidemia, smoking and vascular disease) and modified ATRIA-HSV were calculated. Participants were divided by in-hospital mortality status into two groups: alive and deceased. Ninety-two (22.4%) patients died. Patients in the deceased group were older, predominantly male and had comorbid conditions. CHADS-VASc-HS (adjusted odds ratio [aOR]: 1.31; p = 0.011), m-RCHADS-VASc (aOR: 1.33; p = 0.007), m-ATRIA (aOR: 1.18; p = 0.026), ATRIA-HSV (aOR: 1.18; p = 0.013) and m-ATRIA-HSV (aOR: 1.24; p = 0.001) scores were all associated with in-hospital mortality. m-RCHADS-VASc and modified ATRIA-HSV had the best discriminatory performance. We showed that m-RCHADS-VASc and m-ATRIA-HSV scores were better than the rest in predicting mortality among COVID-19 patients.