• Percentage of patients in the glucose target range within 24 hours of the initiation of treatment. [ Time Frame: 24 hours ]
  

• Modified Rankin Scale at three months [ Time Frame: three months ]
  
• Magnetic resonance imaging (MRI) infarct growth at one day [ Time Frame: one day ]
  
• Percentage of patients with hypoglycaemic event (< 3 mmol/l)
  

It is demonstrated that post-stroke hyperglycaemia is associated with poor outcome. Yet, the efficiency of aggressive insulin serum glucose control is not established. Furthermore the modalities of insulin administration are discussed since US and European guidelines recommend subcutaneous administration, whereas some groups use intravenous administration of insulin. The purpose of this study is to compare the efficacy and the safety of a continuous intravenous insulin protocol versus classical subcutaneous administration of insulin. The study will include carotid territory stroke at the acute stage (< 6 hours of onset) confirmed by MRI, and randomized in intravenous versus subcutaneous insulin. The percentage of patients in the target range (mean capillary glycaemia within 24 hours < 7 mmol/l) is the primary outcome. The secondary outcomes include comparison of Modified Rankin Scale at 3 months, rate of hypoglycaemic events and comparison of Infarct Growth as measured on MRI.

Inclusion Criteria:

• Age over 18 years
• Carotid territory stroke
• MRI performed in the first five hours of stroke onset
• Baseline National Institutes of Health Stroke Scale (NIHSS) > 4 or < 26
• Time between MRI and treatment under one hour

Exclusion Criteria:

• Pre-existing Modified Rankin Scale of three or higher
• Advanced or terminal illness with risk of death in the next 6 months, addiction