The aim of the present study was to determine whether any benefit might occur from lowering the glycaemic index of diet in the medium term in diabetic patients. Eighteen well-controlled diabetic patients (12 Type 1 and 6 Type 2 non-insulin-treated), were assigned to either a high mean glycaemic index or low mean glycaemic index diet for 5 weeks each in a random order using a cross-over design. The two diets were equivalent in terms of nutrient content and total and soluble fibre content. The glycaemic indices were 64 +/- 2 (mean +/- SD) % and 38 +/- 5% for the two diets. The high glycaemic index diet was enriched in bread and potato and the low glycaemic index diet in pasta, rice, and legumes. At the end of the study periods, the following variables were improved on the low compared to the high glycaemic index diet: fructosamine (3.9 +/- 0.9 vs 3.4 +/- 0.4 mmol l-1, p less than 0.05); fasting blood glucose (10.8 +/- 2.8 vs 9.6 +/- 2.7 mmol l-1, p less than 0.02); 2-h postprandial blood glucose (11.6 +/- 2.9 vs 10.3 +/- 2.5 mmol l-1, p less than 0.02); mean daily blood glucose (12.0 +/- 2.5 vs 10.4 +/- 2.7 mmol l-1, p less than 0.02); serum triglycerides (1.5 +/- 0.9 vs 1.2 +/- 0.6 mmol l-1, p less than 0.05). No significant differences were found in body weight, HbA1C, insulin binding to erythrocytes, insulin and drug requirements, and other circulating lipids (cholesterol, HDL-cholesterol, phospholipids, Apolipoprotein A1, Apolipoprotein B). Thus the inclusion of low glycaemic index foods in the diet of diabetic patients may be an additional measure which slightly but favourably influences carbohydrate and lipid metabolism, requires only small changes in nutritional habits and has no known deleterious effects.