Background Vitamin D deficiency has been associated with an increased risk of respiratory infections. Objectives The study aimed to evaluate the serum 25-hydroxyvitamin D [25(OH)D] concentration in patients admitted to the intensive care unit (ICU) as a predictor of coronavirus disease 2019 (COVID-19) mortality. Methods A single-center retrospective observational study was conducted. Forty adult patients (50% men) with confirmed COVID-19 who were admitted to the ICU were enrolled. The primary endpoint was mortality at day 60. Serum 25(OH)D concentration was measured on the day of admission to the ICU. We used the Mann-Whitney test, Fisher's exact test, Kaplan-Meier analysis, and receiver operator characteristic (ROC) analysis to assess serum 25(OH)D concentration as a predictor of COVID-19 mortality. Results All 40 patients had a low median (IQR) serum 25(OH)D concentration at admission [12 (9-15) ng/mL]. The median (IQR) serum 25(OH)D concentration was greater in survivors [13.3 (10.0-17.1) ng/mL, n = 22] than in nonsurvivors [9.6 (7.9-14.2) ng/mL; n = 18], P = 0.044. The area under the ROC curve was 0.69 (95% CI: 0.52, 0.86; P = 0.044). The 60-d mortality rate of those with serum 25(OH)D concentrations ≤9.9 ng/mL (n = 14, 71%) tended to be greater than that of those with concentrations >9.9 ng/mL (n = 26, 31%) (P = 0.065), and they had a 5.6-fold higher risk of death (OR: 5.63; 95% CI: 1.35, 23.45; P = 0.018). Conclusions The ICU patients had a low serum 25(OH)D concentration. Serum 25(OH)D concentrations ≤9.9 ng/mL on admission can be used to predict in-hospital mortality in patients with COVID-19.This trial was registered at clinicaltrials.gov as NCT04450017.