Selection of an appropriate ensemble of General Circulation Models (GCMs) is vital to properly assess the impacts of climate change in a region. Several methods exist for the ranking and selection of GCMs. The present study compares the commonly-used Compromise Programming (CP) method and the recently-developed Global Performance Indicator (GPI) technique for ranking the GCMs. The ability of 20 GCMs to simulate the observed monthly rainfall over the period 1961–2005 in the Upper Godavari River basin in India is assessed, and a grid-wise analysis is performed. The SPAtial EFficiency (SPAEF) metric and Kling–Gupta Efficiency (KGE) are used as the performance indicators to evaluate the 20 GCMs. Further, both the CP method and the GPI technique are applied to integrate the values of the performance indicators into one. The group decision-making (GDM) approach is employed to make a collective decision about the rank of the 20 GCMs considering all the grids. The results from the CP method suggest that the best models are MPI-ESM-P, MPI-ESM-LR, and CNRM-CM5-2, whereas those from the GPI technique indicate that the best-performing GCMs are NorESM1-M, FIO-ESM, and MPI-ESM-LR. An analysis is also performed to examine whether the ranking of the GCMs identified based on the grid-wise analysis also holds true when the average rainfall over the entire basin is considered for ranking. To this end, the average rainfall value across all the 12 grids in the basin is used for the evaluation. The ensemble of GCMs identified by the grid-wise study of GCMs using the CP method provides better SPAEF and KGE values when compared to that using the GPI technique. These results suggest that the GCMs have to be evaluated at the individual grids, and then collective information has to be taken to identify the ensemble of the best-performing GCMs.