Climate response to global warming in China is significant for predicting future climate change risks and formulating adaptation and mitigation policies. This study examined the projections of climate extremes in China under 1.5 and 2 °C global warming based on simulations of 25 climate models from the Coupled Model Intercomparison Project phase 6 under SSP2-4.5 and SSP5-8.5 scenarios. Different reliability ensemble averaging methods were employed to evaluate performance. The results indicated that the upgraded reliability ensemble averaging method showed the best performance in simulating climate indices in China, with the smallest biases compared with observations. There were increases in all temperature and precipitation indices. The increase in the magnitude of the indices under the SSP5-8.5 scenario was slightly greater than that under the SSP2-4.5 scenario. The annual mean temperature, maximum temperature, and minimum temperature, when averaged over the whole of China under the SSP5-8.5 scenario, increased by 1.11, 1.18, and 1.31 °C (1.88, 1.98, and 2.14 °C), respectively, relative to 1995-2014, for 1.5 °C (2 °C) above-preindustrial global warming levels. Increases in Prcptot and R95p were 5.6% and 14.4% (10.5% and 25.7%, respectively). The most remarkable warming occurred in northern China and parts of the Tibetan Plateau. Prcptot and R95p levels increased significantly in most of Western China. Under an additional 0.5 °C of global warming, all temperature indices are expected to increase by more than 0.5°C across China. Prcptot will increase by an additional 4.9%, and R95p will increase by 11.2% when averaged over China under the SSP5-8.5 scenario. Under 2 °C global warming, the probability of all temperature indices increasing by 1.5 °C in China is greater than 50%, except for a few parts of South China. The probability of a Prcptot (R95p) increase of 5% (15%) threshold greater than 50% is found in North China (almost the whole of China). Although the upgraded scheme has reduced the uncertainty of projections to some extent, the development of integration methods and downscaling techniques is required to provide more accurate future projections.