Regular or crystalline insulin with sodium glycocholate as surfactant administered intranasally to normal volunteers induced hypoglycaemia and an increase in serum immunoreactive insulin concentrations. Serum C-peptide concentrations decreased or remained unchanged. Insulin administered intravenously to three of these subjects yielded a potency ratio of 1:8 for intranasal and intravenous insulin. In four insulin-dependent diabetics a cross-over study was performed on different days, insulin being administered once intranasally and once subcutaneously in a ratio of 1:9. In these patients the intranasal insulin was more effective than the subcutaneous insulin in preventing hyperglycaemia after breakfast. In four other insulin-dependent diabetics 11-hours monitoring was performed twice on two different days, insulin being administered in divided dosage sufficient to achieve a reasonable glycaemic profile. The administration during the morning, whereas subcutaneous insulin was more effective than intranasal during the afternoon.