Twenty-one insulin-dependent diabetic children completed a trial of 4 different breakfasts, given in random order. Three diets differed in fibre content. The fourth diet contained soya beans as part (38%) of the dietary fibre source. Children collected capillary blood samples on to filter paper strips which were analysed for blood glucose content. Each morning the children were asked to exercise vigorously for an hour and to rest for an hour, resulting in comparable rest and exercise periods for each child. Mean initial blood glucose levels on the 4 diets were not significantly different. The low-fibre diet resulted in the highest blood glucose concentrations after breakfast. Blood glucose levels on the high-fibre diet did not differ from those on the medium-fibre diet. The bean diet produced the lowest mean blood glucose level and the smallest reduction in blood glucose level in the hour before lunch. All the children found the bean diet unacceptable but liked the high- and medium-fibre diets, which were as popular as the low-fibre diet. The level of prescribed exercise had no effect on the level of blood glucose. It appears that the potentially major benefits from beans are limited by their unpalatability. The more acceptable cereal fibre produces a smaller but important benefit on morning hyperglycaemia after breakfast.