This paper deals with reconstruction of the road grade profile from GPS data. The several repeating driving cycles are recorded on a conventional bus route by using GPS measurement equipment. The road grade is reconstructed by using several filtering- and averaging-based methods, as well as the Gaussian process method which inherently interpolates input data while successfully dealing with the measurement noise. For the sake of assessment of different approaches, the reconstructed road grade profiles are correlated with the original/recorded ones. Further, the model of EV bus is presented and used for simulation-based sensitivity analyses of road grade impact on EV energy consumption. Also, the same EV model is simulated over the recorded driving cycles in order to estimate realistic error introduced by neglecting the road grade within EV-related studies.