In subjects with untreated type 2 diabetes, we previously determined that a weight-maintenance, non-ketogenic diet containing 30 % protein, 50 % fat and 20 % carbohydrate (30:50:20) decreased the percentage total glycohaemoglobin (%tGHb) by 2·2 % glycohaemoglobin over 5 weeks compared to a diet recommended for the American public (protein–fat–carbohydrate 15:30:55). Both the fasting and postprandial glucose were decreased. The objective of the present study was to determine if increasing the carbohydrate content from 20 to 30 % at the expense of fat would still provide a similar effect on %tGHb, fasting and postprandial glucose concentration. Eight men with untreated type 2 diabetes were studied over a 5-week period. Results at the beginning (standard diet) and end of the 5-week study were analysed. Body weight was stable. Fasting glucose concentration decreased by 40 %; 24 h glucose area response decreased by 45 %. Insulin did not change. Mean %tGHb decreased by 1·7 (from 10·8 to 9·1 %), and was still decreasing linearly at 5 weeks. In conclusion, a high-protein, 30 % carbohydrate diet could be a patient-empowering method of improving the hyperglycaemia of type 2 diabetes without pharmacologic intervention. Long-term effects and general applicability of this diet remain to be determined.