An attempt was made to improve the control of blood glucose in patients with insulin-dependent diabetes mellitus. Ten patients were studied over a period exceeding 8 mo in the outpatient department. Each patient was taught to monitor his or her own blood glucose concentration and was enrolled in an exercise program. In addition, they were taught to calibrate insulin, food, and exercise in terms of the response of the blood glucose. After 3 mo of study, there was a significant decrease in hemoglobin A1c (Hb A1c) values from a mean concentration of 10.3 to 7.6%. Hb A1c values correlated well with mean blood glucose levels (r = 0.86). Dietary analysis revealed that the patients selected a diet consisting of 25% protein, 44% fat, and 31% carbohydrate, but there was a wide range in caloric distribution. There was no correlation between blood glucose control as measured by Hb A1c and caloric distribution. Systolic blood pressures decreased greater than 10 mm Hg. Serum alkaline phosphatase also decreased in each patient as control of blood glucose improved. Three of the patients with abnormalities of nerve conduction measurement at the beginning of the study had a complete return to normal at 8 mo. An outpatient program such as this may be cost effective through the avoidance of diabetes-related hospitalizations and may provide a mechanism whereby larger studies can be designed to define further the relationship of glycemia to the sequelae of diabetes.