INTRODUCTION: The purpose of this study was to assess the relation between glycaemic control and the severity of sepsis in a cohort of patients treated with intensive insulin therapy (IIT). METHODS: In a prospective, observational study, all patients in the intensive care unit (ICU) (n = 191) with sepsis, severe sepsis or septic shock were treated with IIT (target blood glucose (BG) level 80 to 140 mg/dl instead of strict normoglycaemia). BG values were analysed by calculating mean values, rate of BG values within different ranges, rate of patients experiencing BG values within different levels and standard deviation (SD) of BG values as an index of glycaemic variability. RESULTS: The number of patients with hypoglycaemia and hyperglycaemia was highly dependent on the severity of sepsis (critical hypoglycaemia < or = 40 mg/dl: sepsis: 2.1%, severe sepsis: 6.0%, septic shock: 11.5%, p = 0.1497; hyperglycaemia: >140 mg/dl: sepsis: 76.6%, severe sepsis: 88.0%, septic shock: 100%, p = 0.0006; >179 mg/dl: sepsis: 55.3%, severe sepsis: 73.5%, septic shock: 88.5%, p = 0.0005; >240 mg/dl: sepsis: 17.0%, severe sepsis: 48.2%, septic shock: 45.9%, p = 0.0011). Multivariate analyses showed a significant association of SD levels with critical hypoglycaemia especially for patients in septic shock (p = 0.0197). In addition, SD levels above 20 mg/dl were associated with a significantly higher mortality rate relative to those with SD levels below 20 mg/dl (24% versus 2.5%, p = 0.0195). CONCLUSIONS: Patients with severe sepsis and septic shock who were given IIT had a high risk of hypoglycaemia and hyperglycaemia. Among these patients even with a higher target BG level, IIT mandates an increased awareness of the occurrence of critical hypoglycaemia, which is related to the severity of the septic episode.