Gene therapy is a potentially excellent approach for the treatment of diabetes instead of pancreas and islet transplantation. However, one difficulty involved in gene therapy for diabetes is a control of insulin/proinsulin production by the cells transfected with insulin cDNA. The purpose of this study is to examine whether control of the proliferation of transfected cells by encapsulation is a feasible approach for the constant delivery of proinsulin to avoid a life-threatening hypoglycemic state. Previously, we established a mouse fibroblast Ltk cells transfected with human insulin cDNA producing human proinsulin (91 ng/24 hr/106 cells). These cells were encapsulated with semipermeable 5% agarose gel and proinsulin production was examined by in vitro long-term culture system. Intraperitoneai implantation into streptozocin (STZ)-induced diabetic mice was performed to investigate in vivo function of the encapsulated cells. The data from the in vitro study demonstrated that encapsulation of 2 x 106 transfectants enabled the stable production of proinsulin for over 80 days (204.4 ± 5.18 ng/ml/day). Implantation of the encapsulated 2 x 107 transfectants improved the hyperglycemic state of diabetic mice for 30 days on the mean value of blood glucose concentration (n = 20). Histological analysis revealed pericapsular inflammation at 30 days after implantation and this may result in malfunction of encapsulated cells. Constant production and delivery of proinsulin could be achieved by encapsulating the human insulin cDNA- transfected cells using 5% agarose. Control of the proliferation of transfected cells appears to be an important factor for constant delivery of human proinsulin toward gene therapy of diabetes mellitus.