Many of the long-term complications seen in insulin-dependent diabetic patients are not averted by conventional modes of insulin therapy. A commonly expressed hypothesis is that more intensive efforts directed at improved metabolic control in diabetic subjects may result in a decrease of long-term complications. Therefore, increasing attention is being devoted to development of strategies designed to achieve improved blood glucose and metabolic control in insulin-dependent diabetic subjects. The approaches can be divided into three categories:  Transplantation,  Mechanical devices, and  Intensive, conventional therapy. The development of the transplantation approach requires resolution of the problem of rejection and the need for immunosuppression. Recent studies indicate that islet transplantation may be performed without long-term immunosuppression. Closed-loop devices for use at the bedside are available at the present time for short-term investigational use. A device for long-term portable use is not available because of the lack of an adequate glucose sensor. Open-loop devices are currently available that enable long-term clinical studies in humans. Some patients have used these devices at Washington University for over two years. In a few, there has been reversal of some of the neuropathy with retinal vascular leakage and near normal, long-term metabolic control. Some findings in over twenty patients studied long-term include the following: 1. Reduction of postprandial hyperglycemia is possible with intravenous. subcutaneous, or intraperitoneal insulin delivery; 2. Hypoglycemia remains the most common problem seen in ambulatory patients treated with open-loop pumps and necessitates glucose monitoring by the patient at home; 3. Long term "normalization" of plasma glucose is not achieved with openloop systems; 4. Maturity onset diabetic patients are easier to control than juvenile onset diabetics; 5. Some patients tolerate hypoglycemia better than others and this may determine who does well long-term; 6. Long-term subcutaneous insulin delivery is safe, whereas other forms of insulin delivery carry a greater risk; 7. Peripheral hyperinsulinemia is frequently found in patients treated by open-loop methods and may have long-term detrimental effects; 8. Behavioral and psychosocial variables may influence long-term results and have not been studied in detail; 9. The primary clinical use of open-loop systems in diabetic patients at the present time is their use in pregnancy and in patients unresponsive to conventional therapy. Our interest in these devices is to determine if they can avert or reverse the long-term complications of the diabetic. We do not consider them safe for routine use in clinical practice and are concerned that practitioners inexperienced in their use may employ these devices prematurely, before their safety and effectiveness are demonstrated in carefully controlled clinical studies.