<p>Global climate change is driving an increase in extreme weather events, resulting in significant losses to human life, ecosystems, and global GDP. Projection of accurate regional climate impacts is essential for developing effective adaptation strategies. However, current global climate models (GCMs) struggle to capture regional extremes due to coarse resolution and oversimplified physical processes. To address these limitations, this study introduces a novel ranking methodology (RM) for selecting GCMs to enhance the accuracy of regional downscaling. Unlike traditional approaches, which focus narrowly on variables like precipitation and 2-m temperature, our method evaluates performance across multiple variables at multiple pressure levels. This comprehensive ranking reduces uncertainties in model selection and improves the ensemble mean performance of GCMs under consideration. Our results identify HadGEM3-GC31-MM, EC-Earth3-Veg, and EC-Earth3-CC as top-performing models, while MCM-UA-1-0 and KIOST-ESM underperform over the selected area of interest. Models like CMCC-CM2-SR5 and NorESM2-MM excel in temperature and precipitation simulations but fail in other critical variables. Similar issues have been identified with other models across multiple variables and pressure level simulations, highlighting the need for holistic model evaluations. The cumulative sum ranking method (Ranking method 3 or RM3), presented in this study, emerged as the most effective methodology, minimizing biases and achieving higher accuracy in precipitation simulations compared to other methodologies. This study presents a refined model selection framework to improve regional climate projections and to provide critical insights for reliable climate impact assessments and adaptation strategies.</p>