<p>Accurate bias correction of climate model rainfall projections is essential for hydrological and climate-impact assessments in regions with complex precipitation regimes, such as the Lake Toba basin in Indonesia. This study evaluates four quantile-mapping (QM) approaches—non-parametric, semi-parametric, parametric, and a Sliding 3-Month Window technique—applied to monthly rainfall from four CMIP6 models across 13 observation stations. Ten probability distributions, including the Alpha-Power Transformed X-Lindley (APTXL), Weibull, GEV, and Gamma, were examined as candidate parametric forms. Although Weibull and GEV provided the best statistical goodness-of-fit to historical rainfall, their performance did not translate into effective bias correction. In contrast, APTXL, while not always the best-fitting distribution, delivered the most consistent improvement across correlation, variability skill, and magnitude-based error metrics within the QM framework. The Sliding 3-Month Window parametric quantile-mapping (PQM)–APTXL approach achieved the highest Comprehensive Rating Index (CRI) among all methods and distributions, reflecting its superior ability to address strong seasonal dynamics and preserve interannual variability. Post-correction assessment showed substantial reductions in MAE and centered RMSE for all CMIP6 models, with EC-Earth3-Veg-LR and the ensemble mean emerging as the most reliable sources of corrected rainfall. These findings highlight the importance of evaluating distributions not only by statistical fit but by functional correction performance, and they demonstrate the effectiveness of APTXL-based quantile mapping for bias correction of tropical rainfall. The study provides a robust basis for improving climate model applicability in hydrological modeling and future climate-impact studies for the Lake Toba region.</p>