While diabetes and admission blood glucose (BG) are established risk factors for adverse outcomes during hospitalization for COVID-19, reports on the impact of prior glycemic control have been variable. We examined the relationship between acute and chronic glycemia on risk of ICU admission, mechanical ventilation (MV) , and mortality among 1,786 patients with diabetes or hyperglycemia (BG > 180 mg/dl twice during any 24-hr period during hospitalization) admitted from March 2020 to February 2021 with COVID-at 5 large university hospitals in the eastern U.S. The cohort was 51.3% male, 53.3% White, 18.8% Black, 29.3% Hispanic, with age = 64.8 ± 13.8 y, BMI = 31.5 ± 7.9 kg/m2, admission BG = 216 ± 134 mg/dl, and HbA1c = 8.1 ± 2.2%. During hospitalization, 38.9% were admitted to the ICU, 22.9% received MV, and 10.6% died. In multivariate regression analysis, among demographic factors, age was the strongest risk factor for in-hospital mortality (OR = 1.per year [95% CI: 1.04, 1.06]) , and Hispanic ethnicity was the greatest risk factor for ICU admission (OR = 1.45 [95% CI: 1.16, 1.80]) and intubation (OR = 1.64 [95% CI: 1.28, 2.10]) . Higher BMI (p = 0.005) and admission BG (p = 0.014) were associated with increased risk of mortality, but HbA1c was not. The glycemic gap (GG) , defined as admission BG minus estimated average BG based on HbA1c, was a stronger predictor of mortality than either admission BG or HbA1c alone. Mortality rate was 5.7% for GG < -20 mg/dl; 12.2% for GG = -20 to < 20 mg/dl; 12.4% for GG = 20 to < 100 mg/dl; and 16.1% for GG ≥ 100 mg/dl (p for trend < 0.001) . Conclusion: Among patients with diabetes or hyperglycemia admitted for COVID-19, in addition to previously established risk factors for poor outcomes (age, Hispanic ethnicity, and BMI) , we found that GG is a stronger predictor of in-hospital mortality than blood glucose alone. This suggests that relative hyperglycemia, as measured by the admission GG, is an important marker of disease severity in COVID-and potentially other serious illnesses. Disclosure M.E.Mcdonnell: Advisory Panel; Everlywell, Inc., Research Support; Lilly, Stock/Shareholder; Abbott Diabetes. G.P.Westcott: None. D.C.Simonson: Stock/Shareholder; GI Windows, Phase V Technologies, Inc. G.Gopalakrishnan: Research Support; Eli Lilly and Company, Spruce Biosciences. R.Garg: None. J.Mitri: Consultant; dairy management, Lnutra. R.S.Weinstock: Research Support; Boehringer Ingelheim International GmbH, Dexcom, Inc., Diasome, Eli Lilly and Company, Insulet Corporation, Kowa Pharmaceuticals America, Inc., Medtronic, Novo Nordisk, Tandem Diabetes Care, Inc., Tolerion, Inc. M.Greenfield: None. N.E.Palermo: Research Support; Dexcom, Inc. R.Radhakrishnan: None. Funding Brigham-TechFoundation, Cambridge, MA 2021