The effects of metformin on glycaemia, insulin and c-peptide levels, hepatic glucose production and insulin sensitivity (using the euglycaemic, hyperinsulinaemic clamp) were evaluated at fortnightly intervals in 9 Type 2 diabetic patients using a stepwise dosing protocol: Stage 1--no metformin for four weeks; stage 2--metformin 500mg mane; stage 3--metformin 500mg thrice daily; stage 4--metformin 1000mg thrice daily. Results are expressed as Mean +/- SEM. Fasting blood glucose decreased from basal values (9.7 +/- 1.0 mmol/L) by 13% at stage 2, 34% at stage 3 and 41% at stage 4 (p less than 0.02 vs basal for all stages; p less than 0.02 stage 2 vs stage 3). Post-prandial glycaemia was significantly improved only with metformin 3000mg/day (p less than 0.05). Fasting, meal-stimulated and total insulin and c-peptide levels showed no change. Hepatic glucose output did not change significantly with metformin. Insulin sensitivity, measured as total glucose utilisation during hyperinsulinaemia, increased from stage 1 (10.3 +/- 2.1 mumoL/kg/min) by 23% at stage 3 (p less than 0.05) and by 29% at stage 4 (p less than 0.02). Basal metabolic clearance of glucose increased compared to stage 1 (1.69 +/- 0.16 mL/kg/min) by 30% at stage 2, 53% at stage 3 and 44% at stage 4 (all p less than 0.02). This study demonstrates that improved efficiency of glucose utilisation, both basally and under conditions of euglycaemic hyperinsulinaemia, is the basis of metformin's antihyperglycaemic action.