<p>This study evaluates the ability of High Resolution Model Intercomparison Project (HighResMIP) models to reproduce seasonal and extreme rainfall over the complex topography of Kerala. It assesses the effectiveness of statistical bias correction. It examines near-future (2026–2050) changes relative to the 1990–2014 period during the southwest monsoon (SWM) and northeast monsoon (NEM) periods, based on daily rainfall observations from 50 IMD stations. The fine-resolution models exhibit moderate improvements in most rainfall features during both seasons compared to their coarse-resolution models. Bias correction using Empirical Quantile Mapping substantially improves the representation of seasonal and extreme rainfall in the historical simulations, while future projections are adjusted using Quantile Delta Mapping. Projected changes are assessed using bias-corrected future simulations relative to corresponding bias-corrected historical simulations. In the near future, rainfall concentration is expected to increase during the SWM, leading to more intense but less frequent rainfall events and longer dry spells. While the seasonal mean remains largely unchanged, interannual variability is projected to increase. In contrast, NEM rainfall is projected to become more evenly distributed, with a higher number of moderate rainfall events and a significant increase in both the mean and interannual variability. SWM extreme rainfall is projected to intensify, with short-duration annual extremes increasing by 7–8% and 100-year return values rising by roughly 34% in the ensemble mean. For NEM, changes in annual rainfall extreme indices are weak and statistically insignificant, and large inter-model spread limits confidence in any robust intensification signal. The changes in SWM seasonal rainfall are largely driven by thermodynamic moistening in response to warming, whereas circulation dynamics primarily govern NEM changes with a weaker thermodynamic influence.</p>