Dynamic traffic assignment methods play a crucial role when estimating a network performance by simulation. Such methods can be decomposed into two parts: definition of the possible routes and determination of the path flow distribution between each OD (Origin/Destination) pair. While the second component has been extensively studied in the literature in connection with network equilibriums (UE, BRUE, SUE, SO...) (Abdulhafedh, 2017; Szeto and Lo, 2005; Peeta and Ziliaskopoulos, 2001), the selection method that reduces the huge number of possible paths in meshed network to a candidate set of paths has received less attention. In particular, to the authors' best knowledge, no extensive sensitivity analysis (SA) has been conducted to assess how such selection methods influence the network performance predicted by simulation.