The circulating insulin and glucose responses in type II diabetic subjects were determined for 5 h after ingestion of various meals, each containing 50 g carbohydrate. The purpose of the study was to 1) systematically study the insulin response to several different high-starch foods, 2) determine whether this insulin response could be predicted by the glucose response, and 3) determine whether the glucose response could be predicted by the physical structure and digestibility of the ingested carbohydrate. Each subject served as his own control. Carbohydrate was given in the form of potatoes, bread, oatmeal, rice, lentils, kidney beans, cornflakes, high-amylose corn muffins, and low-amylose corn muffins. Bread, oatmeal, rice, lentils, kidney beans, and high-amylose corn muffins resulted in a significantly lower glucose area than 50 g glucose, and the glucose response generally could be predicted by the physical structure and the known digestibility of the ingested carbohydrate. The insulin rise was statistically significantly greater than would be predicted from the glucose response for oatmeal, lentils, kidney beans, and high-amylose corn muffins. Although not statistically significant, the mean was greater than predicted for every other food except potatoes when the insulin response to 50 g glucose was used as a standard. These results indicate that the insulin response cannot be predicted by the glucose response.