The aim of this study was to assess the impact of glycaemic control with and without pre-existing diabetes on outcomes in patients with severe COVID-19. This was a prospective observational study of 109 patients with severe COVID-19 treated in either intensive care or high dependency care units at National Hospital Kandy, Sri-Lanka. The most prevalent co-morbidity in severely ill COVID-19 patients was diabetes (66%). hose with HbA1c >/=10% had non-significantly longer duration of hospitalization (median 21 vs. 13 days; p=0.073) than those with HbA1c <10%. Individuals treated with steroids for COVID-19 had higher average capillary blood glucose (CBG) during hospitalization (mean 182.9 mg/dL; SD 47.47 vs. 157.28 mg/dL; SD 34.01; p=0.006), but there was no significant difference before and after commencing steroids (p = 0.593). Daily glycaemic variation showed peak at mid-day. Among 76 who required insulin in-hospital, 46 had non-insulin-dependent diabetes, 15 had no pre-existing diabetes, and 14 were not commenced on steroids. Average daily insulin requirement was higher in those who required oxygen (median 24U vs. 0U; p=0.029). Higher proportion of those who required insulin required intubation-and-ventilation (53.3% vs. 29.2%; p=0.012). Higher proportions of those who developed hypoglycaemia required inotropes (72.4% vs. 37.5%; p=0.001), haemodialysis (24.1% vs. 7.55; p=0.018) and developed high D-dimer (>750ng/mL) (94.7% vs. 70.5%; p=0.033) than those without hypoglycaemia. They had a longer duration of hospital stay (median 16 vs. 11.5 days; p=0.047). Fungal sinusitis developed in 4, of which all had pre-existing diabetes; 2 of whom had HbA1c >10% and were previously insulin dependent. 3 individuals developed persistence of new onset diabetes at 3 months post-COVID-19, of which 1 was not treated with steroids for COVID-19. Pre-existing diabetes status, as well as glycaemic control and fluctuations during the acute COVID-19 infection are associated with acute and long-term COVID-19 related adverse outcomes.