Background: Debate between the Calorie Restriction (CR) and Low-Carbohydrate Diet (LCD) has been continued for several years. We have started LCD therapy for diabetic patients since 1999, and experienced good response to control hyperglycemia. Recently the risk of postprandial hyperglycemia for cardiovascular events is reported, and the range of glucose level is focused. Morbus (M) value is proposed to be a good index to show the range of glucose fluctuation. Usefulness of LCD and usefulness of M-value were shown by combined analysis. Methods: Twenty-six patients with Type 2 Diabetes Mellitus (T2DM), who admitted to the Takao Hospital, were the subjects. They were 12 males and 14 females, and their age ranged from 38 to 78 years old. A few patients were new patients, but mostly referred from other hospitals to receive LCD treatment. All patients started from CR therapy with 60% carbohydrate for 2 days, and then carried out LCD dietary therapy with 12% carbohydrates for 10-12 days. On day 2 and 12, blood glucose level was measured at 7 points a day to see the fluctuation from morning to night. Blood and urinary samples were corrected on the same day to get ordinary biochemical data and metabolic marker. Results: To know the effect of LCD on different level of HbA1c, the patients were grouped to low (L; 6.1±0.5%), middle (M; 7.9±0.4%) and high (H; 9.2±0.7%) by HbA1c value. The M-values after 2 day CR were 26.1 (L), 94.1 (M) and 343 (H), respectively. The LCD therapy decreased M-value to 10.4 (L), 18.8 (M) and 84.2 (H), respectively. The significant reduction of M-values was a reflection of the improved postprandial glucose level. LCD was effective even in the bad controlled diabetic patients. In addition to the decreased triacylglycerol and 24 hr urinary C-peptide, uric acid increased in all patients. Conclusion: Ten to 12 day-dietary therapy with LCD was quite effective to improve blood glucose profile. Patients with high HbA1c could be safely treated by LCD. Blood glucose improvement was well correlated to the M-value.