In order to study the influence of non-carbohydrate foods on responses to carbohydrates during mixed meals, 30 Type 2 (non-insulin-dependent) diabetic patients followed a standardized diet for 5 consecutive days. On days 2, 3, 4, and 5, four different lunches were eaten in a randomized order. Lunches consisted of rice or glucose (50 g carbohydrate) eaten either alone, or as part of a mixed meal (32 g protein, 20 g fat). Glucose and insulin levels prior to the lunches did not differ significantly. Glucose and insulin responses differed (p less than 0.001), responses to rice being lower than responses to glucose, and responses to the meal plus rice lower than responses to the meal plus glucose. The ratios of glucose responses to rice and glucose (glycaemic index) were similar (alone, 47 +/- 4%; meal, 47 +/- 4%; NS). The ratios of the insulin responses did not differ (78 +/- 8 vs 96 +/- 7%; NS). The blood glucose responses to mixed meals were lower than responses to the carbohydrates eaten alone, the ratios being similar (rice, 52 +/- 3%; glucose, 58 +/- 5%; NS). Serum insulin responses were however higher, and the ratios differed (172 +/- 13 vs 138 +/- 14%; p = 0.05). These results suggest that the glycaemic index of isolated carbohydrate foods can predict the relative response to mixed meals in the same individuals with Type 2 diabetes. They also support the insulin secretagogue effect of non-carbohydrate foods, which may vary according to the source of carbohydrate in the meal.