The outputs of General Circulation Models (GCMs), owing to systematic biases and their coarse spatial resolution require bias correction before being applied to local-scale climate change impact assessments, particularly in vulnerable arid regions. However, the performance of various methods in reproducing daily precipitation extremes in these climates has not been extensively evaluated. This research assesses the performance of a comprehensive suite comprising 21 diverse bias correction methods for daily precipitation data derived from two CMIP6 models (BCC-CSM2-MR and CanESM5) across six synoptic stations located in the arid region of southeastern Iran during the period 1989–2014. The results revealed substantial divergence in performance among the methods. The majority of conventional methods, including Empirical Quantile Mapping (EQM), failed primarily because they were unable to adequately handle the high frequency of non-precipitation days, producing excessively large error values. In contrast, three methods—Quantile Ranking Bias Correction (QRBC), Gamma Quantile Mapping (GAMMA), and Gamma-Pareto Quantile Mapping (GPQM)—demonstrated superior performance. Among these, the QRBC method was identified as the most optimal approach, as it was able to accurately reproduce the entire statistical distribution, including the intensity and frequency of daily extreme events. The superiority of these methods was consistent across all stations and for both GCM models, suggesting the robustness and broader applicability of the findings. This study emphasizes the necessity of selecting bias correction methods tailored to the region’s climatic characteristics and introduces QRBC as a reliable tool for future climate change studies in arid regions.