So far no really successful therapy has been found for diabetes mellitus, though it is a common disease and its complications, involving microangiopathy, are extremely serious. The serious limitation of all forms of insulin therapy (various types of insulin and insulin pump) is the impossibility of achieving a fast, successful response to either the variations in blood sugar levels after eating or increased glucogenogenesis caused by glycosteroids and adrenalin. The prolonged hyperglycaemic peaks which result, may well be one of the possible causes of microangiopathy. Only a system which responds to blood sugar variations by stimulating adequate insulin secretion immediately will provide correct prophylaxis for the complications of diabetes mellitus. At the moment only P-pancreatic cells have this capacity and in theory pancreas transplantation seems to be the best treatment for diabetes. However out of 105 complete pancreas transplants only 5 survived for over a year, with the aid of immunosuppressants. Equally none of the 66 allotransplants of pancreatic islands in immunosuppressed patients was successful in the long term. It is also important to bear in mind that long term immunosuppression may prove more dangerous than microangiopathy itself. For these reasons a new cell transplant system using molecular sieve membranes to create an immunological barrier is being investigated. Although it is too early to forecast long-term results, continuous progress is being made and recent discoveries about the physiology of beta pancreatic cells encourage the belief that this approach may be successful in the future.