We review progress towards the goal of utilizing stem cells as a source of engineered pancreatic beta-cells for therapy of diabetes. Protocols for the in vitro differentiation of embryonic stem (ES) cells based on normal developmental cues have generated beta-like cells that produce high levels of insulin, albeit at low efficiency and without full responsiveness to extracellular levels of glucose. Induced pluripotent stem (iPS) cells also can yield insulin-producing cells following similar approaches. An important recent report shows that when transplanted into mice, human ES-derived cells with a phenotype corresponding to pancreatic endoderm matured to yield cells capable of maintaining near-normal regulation of blood sugar [Kroon et al., 2008]. Major hurdles that must be overcome to enable the broad clinical translation of these advances include teratoma formation by ES and iPS cells, and the need for immunosuppressive drugs. Classes of stem cells that can be expanded extensively in culture but do not form teratomas, such as amniotic fluid-derived stem cells and hepatic stem cells, offer possible alternatives for the production of beta-like cells, but further evidence is required to document this potential. Generation of autologous iPS cells should prevent transplant rejection, but may prove prohibitively expensive. Banking strategies to identify small numbers of stem cell lines homozygous for major histocompatibility loci have been proposed to enable beneficial genetic matching that would decrease the need for immunosuppression.