The objective of this study is to explore the association between documented diabetes, fasting plasma glucose (FPG), and the clinical outcomes of Coronavirus disease 2019 (COVID-19). This retrospective study included 255 patients with COVID-19. Of these, 214 were admitted to isolation wards and 41were admitted to intensive care units (ICUs). Demographic, clinical, treatment, and laboratory data were collected and compared between ICU and non-ICU patients. Multivariable logistic regression models were used to explore the risk factors associated with poor clinical outcomes (ICU admission or death). There were significant changes in several clinical parameters in ICU patients (leukopenia, lymphopenia, elevated D-dimer, as well as higher levels of FPG, cardiac troponin, serum ferritin, IL-6, and high-sensitivity C-reactive protein)compared with non-ICU patients. The prevalence of known diabetes was substantially higher in ICU than non-ICU patients (31.7% vs. 17.8%, P = 0.0408). Multivariable regression analysis showed that a history of diabetes [odds ratio (OR), 0.099; 95% confidence interval (CI), 0.016-0.627; P = 0.014], high FPG at admission (OR, 1.587; 95% CI, 1.299-1.939, P < 0.001), high IL-6 (OR, 1.01; 95% CI, 1.002-1.018, P = 0.013), and D-dimer higher than 1 mg/L at admission (OR, 4.341; 95% CI, 1.139-16.547, P = 0.032) were independent predictors of poor outcomes. Cox proportional hazards analysis showed that compared with FPG < 7 mmol/L, FPG levels of 7.0-11.1 mmol/L and ≥ 11.1 mmol/L were associated with an increased hazard ratio (HR) for poor outcome (HR, 5.538 [95% CI, 2.269-13.51] and HR, 11.55 [95% CI, 4.45-29.99], respectively). Hyperglycemia and a history of diabetes on admission predicted poor clinical outcomes in COVID-19. Copyright © 2020 Elsevier B.V. All rights reserved.