The dependency of the results from a network adjustment on the application used is investigated. For that purpose, the results obtained by each tested application on several sets of simulated measurements are compared. In each simulation, only one parameter varies. We first present our comparison methodology and the method that was used to add Gaussian-like errors to theoretical measurements. We then apply it to study the impact of the side length of the network and of the ellipsoidal height difference among points in the network for several network adjustment applications: Columbus, CoMeT, Geolab, JAG3D, LGC, Move3, Star*Net and Trinet+.