In this study, the effects of implementing stochastic finite fault ground motion simulations in earthquake hazard and risk assessment are evaluated. The investigations are conducted for the city of Dehradun (Indian Himalayas). We compare two ground motion estimation techniques: a ground motion prediction equation– based technique and a simulation-based technique. The comparison focuses on the differences the techniques imply on earthquake damage and loss estimates. Ground motion simulations are first calibrated against the instrumental recordings of the 1991 Mw 6.8 Uttarkashi earthquake. Afterward, a number of events are considered with different magnitude, distance, and azimuth to the source. Results indicate large differences between ground motion and loss estimates derived by the two methods, especially in the direction of rupture propagation, which persist to 2–2.5 fault lengths distance. It is therefore strongly recommended to consider rupture kinematics and orientation to the test bed when providing ground motion estimates for near-field earthquake loss assessment studies.