We assessed the efficacy of intranasal aerosolized insulin containing laureth-9 as a surfactant in patients with Type I diabetes by fasting studies in 8 patients, mixed-meal studies in 15, and long-term home use in 8. The intranasal insulin (1 U per kilogram of body weight in 1 per cent laureth-9) was rapidly absorbed (in 15 minutes); it lowered the plasma glucose level by 50 per cent in 45 minutes in fasting normal controls and by 50 per cent in 120 minutes in fasting diabetics. The glucose-lowering potency depended on the insulin dose and surfactant concentration. Nasal irritation was proportional to surfactant concentration, with great variability among subjects. After intranasal insulin used before meals (1 U per kilogram in 1 per cent laureth-9), the two-hour postprandial glucose level increased above before-meal levels by 38 mg per deciliter, as compared with 191 mg per deciliter after intranasal placebo in patients with Type I diabetes (P less than 0.05). An outpatient feasibility study examining three months of use of intranasal aerosolized insulin before meals as a supplement to Ultralente insulin revealed that the aerosol was well tolerated, with glycemic control (as indicated by the percentage of glycohemoglobin, home glucose measurements, and hypoglycemic reactions) comparable to that during a subsequent three-month period of conventional subcutaneous insulin treatment. The results suggest that intranasal insulin has potential as an adjunct to subcutaneous insulin in the therapy of Type I diabetes.