Intensive insulin therapy to control blood glucose has been found to reduce mortality among critically ill patients in a surgical intensive care unit, though a simple prescriptive insulin infusion protocol to achieve this has not been published previously. This study documents the development and routine use of a simple prescriptive intravenous insulin infusion protocol for critically ill patients and compares the results with previous practice. During development the protocol was optimized and practical issues of implementation addressed. The optimized protocol was then used for all ICU admissions, and a prospectively defined retrospective chart audit performed for the first month of use. Results were compared with a similar time period the previous year. In September 2002, 27 admissions were started on the protocol. Blood glucose for the time on the protocol had a median value of 6.2 (IQR 5.9-7.1) mmol/l compared with 9.2 (IQR 8.1-10.2) mmol/l for those on insulin in 2001. Blood glucose for the whole ICU stay for those on the protocol in 2002 had a median value of 6.6 (IQR 6.0-7.4) mmol/l compared with 8.6 (IQR 8.0-9.4) mmol/l in 2001. Blood glucose for all ICU patients in 2002 had a median value of 6.5 (IQR 6.0-7.3) mmol/l compared with 7.2 (IQR 6.3-8.3) mmol/l in 2001. Three blood glucose recordings were less than 2.2 mmol/l in September 2002. This study provides initial effectiveness and safety data for the Bath Insulin Protocol Further audits in a larger patient population are now needed.