To compare the effect of adding metformin to insulin therapy with a moderate increase in insulin dose alone in insulin-treated, poorly controlled Type 2 diabetic patients, 47 consecutive such patients (baseline daily dose >0.5 IU kg(-1) and HbA1c >8%) were openly randomized either to a combination of their previous insulin schedule plus metformin (2.55 g daily in three divided doses, n = 24) or to a moderate insulin dose increase (20% of baseline, n = 23). The patient status/biochemical profile was assessed at entry and at 4 months. Among those assigned to insulin + metformin, 18 took the drug. Upon an intention-to-treat basis, patients assigned to insulin dose increase had a statistically significant weight gain (1.16+/-1.9 vs 0.3+/-4.5 kg, p < 0.05). Patients assigned to the insulin + metformin regimen experienced a significantly greater fall in HbA1c (-1.87+/-2.16 vs 0.03+/-1.68%, p < 0.01), total cholesterol (-0.56+/-0.89 vs 0.14+/-0.72 mmol l(-1), p < 0.05) and LDL-cholesterol (-0.51+/-0.73 vs 0.19+/-0.6 mmol l(-1), p < 0.01). These data suggest that adding metformin to insulin in poorly controlled Type 2 DM patients offers an advantage in terms of glycaemic control and lipid plasma profile.