Twenty-four diabetic patients receiving insulin were randomized to 3 groups. Group I began self blood glucose monitoring by meter and switched to visually read strips at 3 months. Group II began self blood glucose monitoring by visually read strips and switched to meter readings at 3 months. Group III monitored urine glucoses for the 6 months of the trial. Professional interaction time was the same for all patients and each patient was placed on the same insulin delivery scheme (3 shots NPH and/or regular). Mean C-peptide levels 6 min following intravenous glucagon was comparable in all 3 groups. Patients monitoring blood glucose showed a significant decrease in glycosylated hemoglobin values (p less than 0.01) from patients monitoring urine at 6 months of the trial. Patients subjectively felt meters were more accurate than visual strips but both groups I and II showed lower glycosylated hemoglobin levels (p less than 0.02 at 6 months) and sequence analysis revealed no sequence effect. The trial confirms that blood glucose monitoring technologies have advantages over urine monitoring in helping patients achieve improved glucose levels. Patients perform equally well in terms of blood glucose "control" whether visually read strips or meters are used for initial teaching or maintenance if patients are instructed appropriately in each methodology. These findings have economic implications for large scale treatment programs.