In this study, a stochastic point‐source model is used to develop a ground‐motion prediction model for northeastern India (NEI) crustal earthquakes calibrated using region‐specific source, path, and site parameters. The NEI strong ground motion database is too limited to derive an empirical ground‐motion prediction model valid for all magnitudes and distances. In the present study, we simulated 30,000 ground motions to develop the horizontal ground‐motion prediction model for NEI. These simulations are for 1000 earthquakes of moment magnitudes (Mw) 4–8.5 and epicentral distances of 1–300 km. In simulating ground motions, quality factor, stress drop, and kappa derived from available NEI strong‐motion records have been used. The model has been validated by comparing it with the available but sparse database of recorded events. The best fit between the model and observed strong ground motion records indicate that the new model is consistent with the available data and can be used in seismic‐hazard analyses until more data are collected and the model is improved.