Macrosegregation originates from the solute partitioning at the liquid-solid interface and the relative motion between liquid and solid phases during solidification of metal alloys. A suitable macrosegregation model should incorporate solidification thermodynamics, crystal growth kinetics, multiphase computational fluid dynamics, and even thermal-structural mechanics. No current model includes all those phenomena, hence leading to assumptions having to be made. This paper discusses some modeling assumptions regarding the treatment of (1) diffusion kinetics of crystal growth, (2) crystal dendritic morphology and (3) solidification shrinkage. Theoretical analyses based on test calculations were made. We find that some previous models, which over-simplified some of the aspects mentioned above for the purpose of computational efficiency, can only be applied to study macrosegregation qualitatively. They lead to significant error estimations of macrosegregation. When the quantitative accuracy for macrosegregation is of primary importance, the multiphase-based models with an appropriate treatment of these aspects, despite the sophisticated computational requirements, are recommended.