Background: In the context of the coronavirus disease 2019 (COVID-19) pandemic and the continuous appearance of mutant strains, accurate selection of patients to be hospitalized is very important for the efficient use of limited medical resources. Methods: We analyzed the characteristics of an oxygen therapy group in 540 hospitalized COVID-19 patients over 50 years old with mild to moderate severity. Results: 119 patients (22.0%) were in the oxygen therapy group, and the proportion of elderly patients was high. Fever, tachypnea, myalgia, and dyspnea were common in this group. Of these patients, 118 (99.1%) had pneumonia, so remdesivir was administered to 110 (92.4%), and dexamethasone was given to 89 (74.8%). On multivariable logistic regression analysis, age was a risk factor for oxygen therapy with an odds ratio (OR) of 1.20 at a 95% confidence interval (CI) of 1.13-1.26. Tachypnea, dyspnea, low E gene cycle threshold, and increased presepsin, ferritin, and procalcitonin levels were also identified as risk factors for oxygen therapy. In the subsequent receiver operating characteristic curve analysis, the area under the curve of age for the prediction of oxygen therapy was 0.788 (95% CI 0.744-0.831, P<0.001). At a cutoff value of 68.5 years old, age showed a sensitivity of 75.0% and a specificity of 71.9%, which are more predictive for oxygen therapy than presepsin, C-reactive protein (CRP), or ferritin. Conclusion:In patients over 50 years old with mild to moderate COVID-19, age is useful, especially at the cut-off value of 68.5 for predicting the risk of oxygen therapy.