To improve plasma glucose control, we administered insulin via the subcutaneous route in seven ambulatory patients with juvenile diabetes (12 to 17 years of age), using a portable infusion pump at a basal rate with pulse-dose increments before meals. After two to four days, the mean plasma glucose (±1 S.E.) of 94±5 mg per deciliter was markedly lower than when insulin was given by conventional methods in the patients’ usual dose (243±28, P<0.01) or in a total dose equivalent to that administered with the pump (150±15, P<0.01). Maximal fluctuations in plasma glucose were also 50 to 150 mg per deciliter below those observed with conventional treatment (P<0.001). Glycosuria was eliminated in six of seven patients during pump treatment. None of the subjects had hypoglycemia. These results demonstrate that plasma glucose can be lowered to normal in ambulatory patients with brittle juvenile diabetes using a portable, subcutaneous insulin infusion system for two to four days. The feasibility and value of the long-term application of this technic need exploration. (N Engl J Med 300:573–578, 1979) THE relation between blood glucose control and the development of long-term microangiopathic and neuropathic complications in diabetes mellitus remains a subject of controversy.1 2 3 A major problem impeding the resolution of this controversy is the observation that even under optimal conditions, conventional treatment with insulin is associated with wide swings in plasma glucose well outside the range of normal.4,5 In healthy subjects on their usual diet, the plasma glucose concentration generally varies by no more than 40 to 60 mg per deciliter, remaining between 70 and 120 when monitored over 24 hours.4,6 In contrast, in brittle, juvenile-onset diabetes attempts at optimal. © 1979, Massachusetts Medical Society. All rights reserved.