Packed distillation columns are usually designed on the basis of a sequential approach, in which first the Equilibrium Stage (EQ) model is used to quantify the separation effort before the required packing height is determined by HETP values. This approach is often sufficient for process design and proper sizing of the column. However, in certain cases the combination of the EQ model and HETP values results in an inappropriate column design. The design accuracy can be improved using a rate-based (RB) approach. Unfortunately, there is no reliable heuristic available to decide which modeling approach should be chosen. Therefore, the current study investigates the need for rate-based modeling in distillation processes by systematically evaluating the differences between EQ and RB composition trajectories. A criterion based on the Euclidean distance is proposed to evaluate the differences between the EQ and RB trajectories. Then, by means of a selectivity analysis basic criteria for model discrimination are derived and validated experimentally. Generally, for wide boiling mixtures the differences between the EQ and RB trajectories are the most distinct. The differences in the trajectories can be further increased or decreased by component specific mass transfer.