Aim. To study the predictive ability of the NEWS2, 4C Mortality Score, COVID-GRAM and qSOFA scales in predicting clinical outcomes in patients with severe coronavirus disease 2019 (COVID-19) hospitalized in a multidisciplinary hospital.Material and methods. The pilot retrospective cohort study used data from 90 patients (52 — intensive care unit subgroup, 38 — general unit subgroup) with a confirmed diagnosis of COVID-19 hospitalized in the O.M. Filatov City Clinical Hospital № 15 (Moscow) from January to March 2021.Results. The probability of a positive outcome of the disease significantly negatively correlates with the patient’s age (R=-0,514; p=0,0002). The best correlation with the COVID-19 outcome had a 4C Mortality Score (R=0,836; p=0,0001). Logistic regression revealed a significant dependence of the “outcome” and “age” parameters with the greatest accuracy in the form of age subgroups according to the World Health Organization classification with odds ratio (OR) of 4,29 (p=0,0001). As a result of ROC analysis, the best predictive ability of disease outcomes was shown for the 4C Mortality Score (area under curve (AUC)=0,878; 95% confidence interval (CI): 0,782- 0,975 (p=0,00001)) and COVID-GRAM (AUC=0,807; 95% CI: 0,720- 0,895 (p=0,00001)); taking into account the division of patients into age subgroups, optimal predictive tools were obtained: in subgroups 18-44 years old and 45-59 years old — the 4С Mortality Score (AUC=0,892, 95% CI: 0,762-0,980 (p=0,002) and AUC=0,853, 95% CI: 0,784-0,961 (p=0,0014), respectively); in the subgroup 60-74 years old — the COVID-GRAM (AUC=0,833, 95% CI: 0,682-0,990 (p=0,038)); in subgroups 75-90 years and >90 years — NEWS2 (AUC=0,958, 95% CI: 0,807-1,0 (p=0,002) and AUC=0,818, 95% CI: 0,713-0,996 (p=0,006), respectively). ROC analysis showed that the age of 70 years is the threshold value, above which the probability of an unfavorable COVID-19 outcome increases significantly (OR=11,63; 95% CI: 9,72- 12,06 (p=0,0052)).Conclusion. The pilot study showed the significance of predicting the hospitalization outcome of patients with severe COVID-19. The 4C Mortality Score and COVID-GRAM scales had the best predictive accuracy. The specificity and sensitivity of the scores depended on the age of a patient. The age of 70 years was the threshold value at which the risk of an adverse outcome increased significantly. Based on the data obtained, it is planned to study the problem of predicting the disease course, taking into account the severity of COVID-19.