The compelling goal of using in vitro differentiation of stem cells to obtain replacement pancreatic beta cells that are clinically effective in treating diabetes has until now eluded researchers. This difficulty raises the question of whether more effective strategies are available. We propose that the native embryonic pathway leading to the definitive endoderm lineage, and continuing on to the endocrine pancreas, is the one most likely to succeed for the in vitro differentiation of embryonic stem cells. We question however whether gain-of-function approaches involving genes necessary for beta cell development are destined to work effectively, and suggest alternative approaches to identifying conditions sufficient for in vitro beta cell differentiation.