The simulation of the Asian monsoon rainfall and its extreme events with high fidelity remains a challenge even for the present day state-of-the-art models with conventional treatment of convection. A multi-scale approach vis-a-vis the super-parameterization appears to overcome the uncertainty of convective parameterization and thereby improve models ability to simulate rainfall. In this study, performance of super-parameterized community climate system model’s atmospheric only (SPCAM) forced with observed SST and coupled (SPCCSM) versions have been evaluated to capture Indian summer monsoon rainfall characteristics. Analyses show that, simulation of rainfall and its extremes are better represented in the atmospheric model (SPCAM) over the Indian landmass. This is largely because of better representation of convection in the uncoupled version. It is also observed that 2–10 day synoptic mode of the summer monsoon has a large variance over Indian region which may be broadly responsible for extreme events, and SPCAM captures this synoptic variability reasonably well. Our study also indicates that models may have poor moisture holding capacity. This problem is more prominent in SPCCSM.