The present study aimed to evaluate the performance of 46 global climate models (GCMs) from the newly released Coupled Model Intercomparison Project Phase 6 (CMIP6) in the historical simulation of precipitation and temperature, and select the best performing GCMs for future projection across China and three major river basins. This study uses four shared socioeconomic pathways (SSPs), namely SSP1‐2.6, SSP2‐4.5, SSP3‐7.0, and SSP5‐8.5 relative to the base period (1961–2014). Initially, 46 models were evaluated across China employing an improved Taylor diagram method. Based on relatively better performance, 10 best‐performing models (TBMs) were selected out of 46 models for further evaluation. The results show that historical temperature was well reproduced by CMIP6 over the study regions with a high correlation coefficient (CC). All the TBMs produced good CC ranging from 0.8 to 0.99 presenting the precipitation and temperature distribution well. Meanwhile, EC‐Earth3 and EC‐Earth3‐Veg well simulated the precipitation and temperature amounts as well as trends over selected three river basins. The multimodel ensemble mean (MME) underestimates temperature over China and selected three basins with bias values of −0.53, −0.21, −0.91, and −0.68°C, respectively. In contrast, MEM overestimated the simulated precipitation with the amount of 27.7, 32.4, 21.0, and 104.6% across China and selected three basins. During future projections, increased precipitation and temperature trends are projected over three selected river basins as well as all across China. The increasing trend of future precipitation over China under SSP1‐2.6, SSP2‐4.5, SSP3‐7.0, and SSP5‐8.5 scenarios are 0.65, 0.86, 1.29, and 0.76 mm·a−1, whereas, the increasing trend of temperature is 0.008, 0.028, 0.050, and 0.065°C·a−1, respectively. Comparatively, the greater the radiation force, the higher projected increases in precipitation and temperature across China and three major river basins were observed. The extent of CMIP6 models over the target region and its river basins calls for further deep assessment of the attribution and possible implementation of robust methods that can accurately simulate the observed patterns for future practice.

South Asia is home to one of the fastest-growing populations in Asia, and human activities are leaving indelible marks on the land surface. Yet the likelihood of successive observed droughts in South Asia (SA) and its four subregions (R-1: semi-arid, R-2: arid, R-3: subtropical wet, and R-4: tropical wet and dry) remains poorly understood. Using the state-of-the-art self-calibrated Palmer Drought Severity Index (scPDSI), we examined the impact of different natural ocean variability modes on the evolution, severity, and magnitude of observed droughts across the four subregions that have distinct precipitation seasonality and cover key breadbaskets and highly vulnerable populations. The study revealed that dryness had significantly increased in R-1, R-2, and R-4 during 1981–2020. Temporal analysis revealed an increase in drought intensity for R-1 and R-4 since the 2000s, while a mixed behavior was observed in R-2 and R-3. Moreover, most of the sub-regions witnessed a substantial upsurge in annual precipitation, but a significant decrease in vapor pressure deficit (VPD) during 1981–2020. The increase in precipitation and the decline in VPD partially contributed to a significant rise in Normalized Difference Vegetation Index (NDVI) and a decrease in dryness. In contrast, a strong positive correlation was found between drought index and precipitation, and NDVI across R-1, R-2, and R-4, whereas temperature and VPD exhibited a negative correlation over these regions. No obvious link was detected with El-Niño Southern Oscillation (ENSO) events, or Indian Ocean Dipole (IOD) and drought evolution, as explored for certain regions of SA. The findings showed the possibility that the precipitation changes over these regions had an insignificant relationship with ENSO, IOD, and drought onset. Thus, the study results highlight the need for considering interactions within the longer climate system in describing observed drought risks rather than aiming at drivers from an individual perspective.