Background: The Veterans Health Administration COVID-19 (VACO) Index incorporates age, sex, and pre-existing comorbidity diagnoses readily available in the electronic health record (EHR) to predict 30-day all-cause mortality in both inpatients and outpatients infected with SARS-CoV-2. We examined the performance of the Index using data from Yale New Haven Hospital (YNHH) and national Medicare data overall, over time, and within important patient subgroups. Methods and findings: With measures and weights previously derived and validated in a national Veterans Healthcare Administration (VA) sample, we evaluated the accuracy of the VACO Index for estimating inpatient (YNHH) and both inpatient and outpatient mortality (Medicare) using area under the receiver operating characteristic curve (AUC) and comparisons of predicted versus observed mortality by decile (calibration plots). The VACO Index demonstrated similar discrimination and calibration in both settings, over time, and among important patient subgroups including women, Blacks, Hispanics, Asians, and Native Americans. In sensitivity analyses, we allowed component variables to be re-weighted in the validation datasets and found that weights were largely consistent with those determined in VA data. Supplementing the VACO Index with body mass index and race/ethnicity had no effect on discrimination. Conclusion: Among COVID-19 positive individuals, the VACO Index accurately estimates risk of short-term mortality among a wide variety of patients. While it modestly over-estimates risk in recent intervals, the Index consistently identifies those at greatest relative risk. The VACO Index could identify individuals who should continue practicing social distancing, help determine who should be prioritized for vaccination, and among outpatients who test positive for SARS-CoV-2, indicate who should receive greater clinical attention or monoclonal antibodies.