Aims/hypothesis. To determine causes of weight gain during insulin therapy with and without metformin in Type II (non-insulin-dependent) diabetes mellitus. Methods. Twenty-six patients with Type II diabetes (body mass index 28 ± 1 kg/m²) were treated with insulin alone (n = 13) or insulin and with metformin (n = 13). Components of energy balance (basal metabolic rate, energy intake, glucosuria) were measured at 0 and 12 months. Results. Glycaemic control improved similarly in patients using (HbA_{1 c} 10.5 ± 0.3 vs 7.6 ± 0.2 %, p < 0.001) and not using (10.2 ± 0.3 vs 7.8 ± 0.3 %, p < 0.001) metformin. The metformin group required 47 % less insulin than the group not using metformin (p < 0.001). Body weight increased by 3.8 ± 0.8 and 7.5 ± 1.6 kg (p < 0.05), respectively. Basal metabolic rate and glucosuria were similar at 0 and 12 months in both groups but the metformin group decreased energy intake by 1.12 ± 0.46 MJ/day, whereas it remained unchanged in the other group (0.15 ± 0.42 MJ/day). Changes in body weight and glycaemia were statistically significant independent determinants of basal metabolic rate. Their change in opposite directions explained why basal metabolic rate remained unchanged. Conclusion/interpretation. Improved glycaemia promotes weight gain by decreasing both basal metabolic rate and glucosuria. Use of metformin decreases weight gain by reducing energy intake and is therefore a useful adjunct to insulin therapy in patients with Type II diabetes. [Diabetologia (1999) 42: 406–412]