To evaluate whether the same amount of carbohydrate (CHO) in different foods gives different glycaemic responses when consumed in the context of a real meal, seven diabetic patients were given in a random order and on alternate days three test meals of identical composition. Each meal consisted of a fixed part to which a 50 g CHO portion of either bread (90 g) or spaghetti (65 g) or potatoes (285 g) was added. The glycaemic response was significantly higher after ingestion of bread than after the spaghetti meal both at 2 (5.9 +/- 0.8 vs 4.3 +/- 0.7 mmol/l X hour, p less than 0.05) and at 5 hours (16.5 +/- 3.6 vs 9.8 +/- 2.3 mmol/l X hour, p less than 0.05). The glycaemic response to the potato meal was similar to that for bread at 2 hours (6.2 +/- 1.2 mmol/l X hour, p less than 0.05 vs spaghetti) and intermediate between the two other test meals at 5 hours (14.6 +/- 4.3 mmol/l X hour). Meal planning for diabetic patients should be based not only on the biochemical properties but also on the glycaemic response to the food.