The postprandial insulin requirements after three mixed meals of equal carbohydrate and energy content were assessed in 10 type-1 and 12 type-2 diabetics by a glucose-controlled insulin infusion system. These were compared with the glycemic response to the same meals of 10 healthy individuals (glycemic index). In type-1 diabetics, we found the highest insulin requirements after consumption of a continental breakfast (low fibre, low protein, high fat). Ten percent less insulin was infused after milk (low fat, high protein) and 30% less after an English breakfast (high fibre, high protein). Type-2 diabetics showed no significant differences in insulin requirements between the three test meals. The glycemic response in healthy individuals had no relation to these insulin requirements. Continental and English breakfast had a similar glycemic effect, whereas milk produced only 30% of the blood glucose response observed after the continental breakfast. These results indicate that neither the carbohydrate content (exchange lists) nor the glycemic index enable prediction of postprandial insulin requirements in insulin-deficient diabetes. For this purpose, we propose the insulin-need index, elaborated by testing whole meals in closed-loop experiments with type-1 diabetics.