BACKGROUND: High-dose glucose-insulin-potassium infusion (GIK) has been suggested to be beneficial in acute myocardial infarction (MI). Recently new large trials have shown no effect of GIK on mortality. To investigate whether metabolic derangement could have negated the potential beneficial effect, we studied the relation between systemic glucose and potassium levels and outcome. METHODS: Patients with signs and symptoms of ST-segment-elevation MI and treated with primary percutaneous coronary intervention (PCI) were randomised to no infusion or high-dose GIK, i.e. 80 mmol potassium chloride in 500 ml 20% glucose at a rate of 3 ml/kg/hour and 50 units short-acting insulin in 50 ml 0.9% sodium chloride for 12 hours. RESULTS: A total of 6991 glucose values and 7198 potassium values were obtained in 476 GIK patients and 464 controls. Mean serum glucose was significantly higher in the GIK group (9.3±4.5 mmol/l vs. 8.4±2.9 mmol/l, p<0.001). Mean potassium level was significantly higher in the GIK group (4.2±0.5 mmol/l vs. 3.9±0.4 mmol/l, p<0.001). Incidence of hyperglycaemia (glucose >11.0 mmol/l) occurred in 70.8% of GIK patients and 33.8% of controls (p<0.001). Hypokalaemia was less common in the GIK group (23.5 vs. 41.2%, p<0.001). Incidence of hyperkalaemia and hypoglycaemia did not differ significantly between the two groups. In multivariate analysis age, previous cardiovascular disease, Killip class >1, unsuccessful PCI and mean glucose after admission were associated with increased one-year mortality. CONCLUSION: In ST-segment-elevation MI patients treated with primary PCI, high-dose GIK induced hyperglycaemia and prevented hypokalaemia. Derangement of the glucose metabolism was related to one-year mortality.