Pancreatic endocrine cells are produced from a dynamic epithelium in a process that, as in any developing organ, is driven by interacting programs of spatiotemporally regulated intercellular signals and autonomous gene regulatory networks. These algorithms work to push progenitors and their transitional intermediates through a series of railroad-station-like switching decisions to regulate flux along specific differentiation tracks. Extensive research on pancreas organogenesis over the last 20 years, greatly spurred by the potential to restore functional β-cell mass in diabetic patients by transplantation therapy, is advancing our knowledge of how endocrine lineage bias is established and allocation is promoted. The field is working towards the goal of generating a detailed blueprint of how heterogeneous cell populations interact and respond to each other, and other influences such as the extracellular matrix, to move into progressively refined and mature cell states. Here, we highlight how signaling codes and transcriptional networks might determine endocrine lineage within a complex and dynamic architecture, based largely on studies in the mouse. The process begins with the designation of multipotent progenitor cells (MPC) to pancreatic buds that subsequently move through a newly proposed period involving epithelial plexus formation-remodeling, and ends with formation of clustered endocrine islets connected to the vascular and peripheral nervous systems. Developing this knowledge base, and increasing the emphasis on direct comparisons between mouse and human, will yield a more complete and focused picture of pancreas development, and thereby inform β-cell-directed differentiation from human embryonic stem or induced pluripotent stem cells (hESC, iPSC). Additionally, a deeper understanding may provide surprising therapeutic angles by defining conditions that allow the controllable reprogramming of endodermal or pancreatic cell populations. Copyright © 2012 Elsevier Ltd. All rights reserved.