In theories of the minor phase (domain) formation in polyblends rendered as emulsions it is usually assumed that the size and shape of the domains are the result of melt viscosity effects (Taylor, Wu) or viscoelasticity effects (VanOene, Elmendorp) being balanced by interfacial tension. This assumption would predict a monotonic decrease of the domain size to a final limiting size with increasing energy of mixing. However, a systematic study of the dependence of domain morphology on industrial mixing processes which was carried out on a "model" LDPE/PS (2/1) mixture and the related polyalloy (i.e., the same mixture with a corresponding block copolymer as compatibilizer) does not support this expectation. Domain size was found to go through a minimum a s mixing energy was increased. A similar minimum was seen in data on specific volume of the melt vs. mixing energy, which indicates a correlation between melt specific volume and domain size. Calculation of the approximate surface area of the domains using a simple model of domain shape indicated that total interfacial energy in the polyblend and/or polyalloy is a trivial part of the mixing energy introduced. These calculations also indicated that if compatibilizer was located entirely at the interface, the surface layer would have a thickness of about 90 nm. Some micrographs seem to show such a surface layer. We propose that a n abrasion mechanism is responsible for the early stage of the dispersion process, and that the final domain size may be controlled by a dispersion-coalescence equilibrium. This is compared with the theories of final particle size proposed by VanOene and Wu. A. P. Plochocki, S. S. Dagli, a n d R. D. A n d r e w s groups the larger domains in the vicinity of the extrudate courtesy of M . J . Doyle, Exxon Corp. center (right hand side of the photograph) (cf. 31 b).