This randomized, open-label, cross-over study compares the efficacy of mealtime rapid-acting analog insulin aspart with human insulin, in combination with metformin. A total of 30 patients with type 2 diabetes, inadequately controlled (HbA(1c)>7.5%) with oral hypoglycemic agents (OHAs), were assigned to human insulin 30 min before meals or aspart immediately before meals, both with metformin 500 mg t.i.d. for 90 days. Patients then switched to the alternate insulin. At 90 and 180 days, blood glucose and lipids were measured at baseline and every 30 min after test meals, for 3h. HbA(1c) and hypoglycemic events were also assessed. After 3 months, HbA(1c) was significantly reduced with aspart, but not human insulin (-0.4+/-0.7% versus +0.1+/-0.7%, p<0.05). During meal tests, blood glucose area under the curve (AUC) was significantly lower with aspart than human insulin (1240+/-476 min/mmol/l versus 1588+/-766 min/mmol/l, p<0.01). AUCs for lipids were similar for both treatments. Neither group experienced serious hypoglycemic events. These results encourage treatment with mealtime insulin aspart plus metformin, in type 2 diabetes patients with postprandial hyperglycemia inadequately controlled by OHAs alone.