Context. Hyper- and hypoglycemia are associated with increased mortality of critically ill patients, but whether this association is causal remains unclear. Early randomized-controlled studies compared insulin infusion targeting "age-normal" blood glucose levels, labeled intensive insulin therapy, with an approach that considered hyperglycemia as a beneficial adaptation. These studies found benefits with maintaining normoglycemia. A recent large multicenter study, NICE-SUGAR, compared a similar "age-normal" with an intermediate glucose target and found the intermediate target superior. These results require explanation.

Evidence acquisition. All published randomized controlled studies on glucose control in ICU were reviewed. The methodological differences between the repeat studies, most specifically NICE-SUGAR, and the original proof-of-concept studies, were systematically analyzed.

Evidence synthesis. There were important methodological differences possibly explaining different outcomes. These comprised different target ranges for blood glucose in control and intervention groups, different routes for insulin administration and types of infusion-pumps, different sampling sites and different accuracies of glucometers, as well as different nutritional strategies and varying levels of expertise.

Conclusions. These differences do not permit confident recommendations for a single optimal glucose target in variable ICU settings. Respecting the "primum non nocere" principle, it appears safe not to embark on targeting "age-normal" levels in ICUs that are not equipped to accurately and frequently measure blood glucose and have not acquired extensive experience with intravenous insulin administration using a customized guideline. A simple overall fall-back position could be to control blood glucose levels as close to normal as possible without evoking unacceptable fluctuations, hypoglycemia and hypokalemia.