The Himalayan region is one of the most seismically active regions in the world. Due to various seismic gaps, there is a possibility of the occurrence of a large seismic event in the region and the probability of such occurrence is being studied by various authors. Highly populated areas are considered areas with higher exposure, where large-scale destruction of properties and loss of lives may occur due to seismic activity, and require region-specific seismic hazard analyses. In this aspect, strong-motion prediction becomes a crucial tool for the prediction of parameters of future earthquakes. In this study, an Artificial Neural Network (ANN) based approach has been used for the task of developing a ground motion prediction model for the prediction of Peak Ground Acceleration (PGA) using independent variables such as earthquake magnitude, hypocentral distance, and site class of the recording station as input data to include the effects of the source, path-propagation, and local site conditions. The dataset is split into 70% for training using the Feed-Forward Back propagation (FFBP) technique and 30% for testing and validation. The data set comprises 152 strong-motion records of 60 earthquake events (Mw >  = 4.0) recorded at multiple stations across the Indian Himalayan range. The authenticity and reliability of the ANN-based prediction model are checked by using the same data set again to predict PGA by developing a Ground Motion Prediction Equation (GMPE) by employing a Two-Stage Maximum Likelihood Regression Analysis (TSMLRA) method. The results from both approaches are evaluated using various parameters, and found that the ANN model has performed relatively better, and it can establish crucial effects such as magnitude saturation, the indication of non-linearity of soil, decay of PGA with distance as a function of magnitude inconsistent with the theory of physical processes of strong motion.