In this paper we examined the relationship between blood glucose control and initiation and progression of increased urinary albumin excretion. In this seven year Oslo study 45 insulin-dependent diabetes mellitus patients were initially randomized into three different modes of treatment: continuous subcutaneous insulin infusion (CSII), multiple injections, or two injections a day. After four years, the patients were free to choose their treatment, and therefore the data were analyzed according to mean HbA1 during seven years. The mean HbA1 was 11.2% (2.2) (SD) at start, and 9.5% (1.5) at seven years, which was a significant long-term improvement (P less than 0.001). Eight out of 10 patients with mean seven year HbA1 less than 8.5% improved their albumin excretion rate, and patients with mean HbA1 greater than 10% had an increased albumin excretion rate (from 26 mg/24 hr to 91 mg/24 hr, P less than 0.02). The glomerular filtration rate decreased slightly regardless of mean HbA1 level. Systolic blood pressure increased significantly regardless of mean HbA1. Diastolic blood pressure was unchanged in the patients with mean HbA1 less than 10%, but increased slightly (NS) in patients with HbA1 greater than 10%. We conclude that mean blood glucose as measured by glycosylated hemoglobin is a main determinant in the progression of urinary albumin excretion in insulin dependent diabetics, and near normoglycemia improves urinary albumin excretion.