Efficacy and reproducibility of insulin administered intranasally as an insulin-deoxycholate 1% (w/v) aerosol to normal and diabetic subjects were assessed by measurements of blood glucose and serum insulin levels. Following administration of 0.5 U insulin/kg with the unconjugated bile salt to fasting volunteers (N = 29), peak serum insulin levels of 103 +/- 49 microU/ml above baseline were observed at 10 min. Blood glucose concentration began to fall by 10 min, reaching 54 +/- 14% of control levels by 30 min, and returning to baseline by 60-80 min. Blood glucose response and peak serum insulin levels were reproducible when the same aerosol dose was repeatedly administered to the same subjects; however, intersubject variations were noted. By comparing serum insulin levels after i.v. and nasal routes of administration, nasal insulin absorption was approximately 10% as efficient as intravenous insulin. Dose response studies revealed that peak serum insulin concentrations were a linear function of the administered dose. In subjects with type I and type II diabetes mellitus, serum insulin levels increased in a manner similar to controls, and resulted in a prompt reduction of blood glucose concentration. However, in contrast to normal subjects, the duration of the glucose response was more prolonged, lasting as long as 5 h. Nasal administration of insulin as an aerosol with bile salts or bile salt analogs should be further evaluated as a possible nonparenteral approach to insulin therapy.