There is considerable uncertainty as to the appropriate timing of preprandial injections of short acting insulin before consumption of a meal. 8 healthy male volunteers were subjected to a glucose clamp (glucose 4.2 mmol/l) by means of a continuous insulin infusion on two occasions. On each occasion the subjects ingested a standardized meal. Twelve units of short acting insulin were injected subcutaneously either immediately before (A) or 30 min prior to the meal (B). In a control study (C) only the meal was consumed. The maximal increment of insulin concentrations after sc injection was 53.1 (6.9) mU/l in A and 58.6 (5.1) mU/l in B and thus comparable to C (52.8 (4.7) mU/l), however it was reached 30 to 60 min later as compared to endogenous stimulation. Injection 30 min prior to meal resulted in higher insulin concentrations early after the meal, but the glucose infusion rate doubled to 5.3 (0.3) mg/kg/min before the meal in order to prevent hypoglycaemia. Omission of this time interval is less hazardous, but bears the risk of late postprandial hyperinsulinaemia.