Before insulin therapy is begun, patients with diabetes are often admitted to hospital. In a retrospective study we have reviewed the initiation of insulin therapy in 54 unselected outpatients (12 of whom were insulin-dependent), when the initial stabilization of therapy was performed predominantly by nurse educators. Most patients found the procedure satisfactory; only one subject indicated dissatisfaction with the regimen and only two indicated that they would have preferred admission to hospital. No patient experienced an acute hypoglycaemic or hyperglycaemic problem that required admission to hospital nor was emergency intervention required during the 12 months that followed the initial stabilization period of insulin therapy. Metabolic control, as measured by glycosylated haemoglobin levels, improved in the majority of both insulin-dependent and non-insulin-dependent patients after 12 months of insulin therapy. A retrospective cost analysis that compared the cost of the outpatient procedure with the cost (hospital-bed costs only) of initiating insulin therapy in a similar group of patients who were admitted to hospital, indicated a saving of $1857 for each outpatient. We conclude that the outpatient initiation of insulin therapy is feasible where the facilities for education about diabetes exist, that it is safe, achieves satisfactory metabolic control, is acceptable to most patients, and offers a considerable saving in costs.