Transitory increase in blood glucose concentration (hyperglycemia) is a very common feature of critical illness and severe trauma. Despite its transitory nature this hyperglycemia is associated with adverse effect and there is a consensus that critically ill patients, irrespective of their diabetic status, benefit from intensive insulin therapy aimed at normalizing blood glucose. The protocols used in intensive care units require that blood glucose concentration be closely monitored and insulin dose titrated or insulin withdrawn in the light of this monitoring. Experience has confirmed that maintaining blood glucose within a target range is challenging and all protocols are associated with risk of continuing hyperglycemia if insulin dose is insufficient, and life-threatening hypoglycemia if insulin dose is too high. It remains unclear whether protocols delivered by a traditional physician-directed model is any more successful at maintaining patient blood glucose concentration within a safe target range than a model that allows nurse autonomy in deciding both insulin dose and frequency of blood glucose measurement. This is the issue addressed in a recently published clinical study conducted at an intensive care unit in Queensland, Australia. For the study 33 patients were assigned to a "physician-directed" group and 37 to a "nursing" group. The hospital protocol demanded that insulin be started in all patients if blood glucose exceeded 10.0 mmol/L. The principal study end point was quality of blood glucose control assessed by two measures: overall mean blood glucose concentration, and % of blood glucose readings above 10.0 mmol/L. Mean (SD) blood glucose concentration of patients in the physician-led group was 8.78 ± 2.74 mmol/L, this compared with 8.33 ± 2.34 mmol/L for the nursing group. Twenty-seven percent (27 %) of blood glucose readings were >10.0 mmol/L (i.e. outside the prescribed target range) in the physician-directed group compared with only 18 % in the nursing group. By these two measures patients in the nursing group achieved markedly better glucose control than those in the physician-directed group. There was no significant difference between the two groups for all secondary end points measured which were: incidence of hypoglycemia, number of blood glucose measurements per 24 hours, median insulin dose/hour and patient survival to discharge. The authors of this small study conclude that glucose-insulin protocols controlled by nurses is feasible, safe and likely to result in better adherence to blood glucose target range. In discussion of their findings the authors speculate that it is the immediate availability of nurses at the patients bedside that explains their findings. The time from finding abnormal blood glucose to change in insulin dose (the decision cycle time) is reduced if nurses at the bedside have autonomy.