Islet transplantation that aims to restore the insulin-producing capacity of the patient has become an attractive approach for the treatment of severe diabetes, ye, perhaps the most challenging problem facing clinical islet implantation is the inexorable loss of islet function in the first 5 years post-implantation leading to only 10% of recipients being insulin-free. This loss of function is not completely understood but believed to be caused by a combination of insufficient implanted β-cell mass, allograft rejection, recurrent autoimmunity, incompatibility of the islet implantation site, and immunosuppressive regiments that are toxic to the islet grafts.   Currently, a large number of isolated human islets are transplanted in severe cases of diabetes through the portal vein of the patient, resulting in partial integration of the islets within the liver. This procedure suffers from high percentage of islet lost, in part, due to lack of a proper microenvironment that is bound to be required for islet viability and function.  Therefore, there is a need to develop in vitro natural three-dimensional structures that mimic the islet tissue microenvironment prior to transplantation. We here describe the preparation of engineered micro-pancreata (EMPs) that are made up of acellular organ-derived micro-scaffolds seeded with human intact or enzymatically dissociated islets.  We show that EMPs secrete quantities of insulin per cell similar to fresh human islets for more than three months in vitro in a glucose-regulated manner. Notably, EMPs respond, even after long periods in culture, by secreting increased amounts of insulin when exposed to Exendin-4, Forskolin, Tolbutamide, Glucagon-like peptide-1 (GLP-1) and Gastric inhibitory polypeptide (GIP).  Quantitatively the amount of insulin and Pdx-1 gene copies per cell are similar to those transcribed by fresh normal human islets.  We also report that when implanted subcutaneously onto Streptozotocin-treated hyperglycemic NOD-SCID the EMPs become vascularized and can rescue the mice in a dose response manner, with a total of 40 human Ieq on EMPs being sufficient for obtaining normoglycemia. Specificity of action of the EMPS was established by removal of the EMPs after 35 days and demonstrating reversal back to the hyperglycemic state. EMPs could thus form the bases for both an assay for “in vivo”-like human b-cell function in the laboratory and as a future treatment modality for cell transplantation in diabetes.