A statistical study was made of measurements within 3 h of injury on 533 patients grouped by injury severity using ISS. A scoring system was also used that took account of the number of injuries. Widely accepted hypotheses about the development of hyperglycaemia were not supported. There was evidence of inhibition of glucose metabolism ('insulin resistance'), but none for any particular mechanism. The only factor that was closely related within any group to [glucose] (plasma glucose concentration) was [lactate], the higher mean value and variance of which after very severe injuries (ISS above 15) could account for much of the higher mean and variance of glucose in this ISS range. At ISS 9-14, up to 1.4 h after injury, only 4% of the variance of glucose could be accounted for by any combination of the concentrations of catecholamines, cortisol, lactate and vasopressin, times of measurement after injury and food intake, and injury severity and number of injuries. There was also no dependence on the part of the body injured. Injury increased the variability of [insulin] less than published statements imply. The increase found was entirely explicable, as expected, by the changes of [glucose] and [adrenaline].