This review describes the three classical models (mosaic, positional, and polarization) proposed to explain blastocyst formation and summarizes the evidence concerning them. It concludes that the polarization model incorporates elements of the other two models and best explains most known information. I discuss key requirements of a molecular basis for the generation and stabilization of polarity and identify ezrin/E-cadherin, PAR proteins, and Cdx2 as plausible key molecular players. I also discuss the idea of a network process operating to build cell allocations progressively into committed differences. Finally, this review critically considers the possibility of developmental information being encoded within the oocyte and zygote. No final decision can be reached on a mechanism of action underlying any encoded information, but a cell interaction process model is preferred over one that relies solely on differential inheritance.