The aim of this study was to determine whether Type 2 diabetic patients should be hospitalised to start insulin therapy. The same structured diabetes treatment and teaching programme (DTTP) was used in outpatients in 10 private practices after postgraduate training of physicians and teaching staff as well as in the Diabetes Department of the University Hospital of Jena, Germany. Seventy-two consecutive Type 2 diabetic patients (ambulatory group) participated in the outpatient DTTP. After one year, 70 patients were re-evaluated and compared with 70 other patients (matched pairs) who were referred to the University Hospital of Jena to start insulin treatment and participated in the same programme during hospitalisation. Initially there were no significant differences between the groups for age, gender, HbA1c, body mass index (BMI), and the time since diagnosis of diabetes. HbA1c (mean normal value 5%) decreased in both groups within the 12 months of follow-up (ambulatory group from 10.3 +/- 2.2 to 8.1 +/- 1.7, p < 0.0001; inpatient group from 10.4 +/- 1.6 to 8.4 +/- 1.7, P = 0.0001). At follow-up there were no significant differences between the groups concerning insulin dosage, HbA1c, severe hypoglycaemia, BMI, and hospitalisation. In combination with a DTTP, the initiation of insulin therapy on an ambulatory basis was as safe and effective as in the inpatient programme. Cost-benefit analysis demonstrated substantial savings in direct costs in the ambulatory programme.