Precipitation extremes have large influences on hydrological, ecological and agricultural systems. This study aimed to investigate variations of precipitation extremes in China's mainland over 1961–2100 based on 27 newly released General Circulation Models (GCMs) in Coupled Model Intercomparison Project phase 6 (CMIP6) under four shared socioeconomic pathway scenarios (SSPs). 10 extreme precipitation indices (EPIs) were extracted for quantifying changes of precipitation extremes. A statistical downscaling method was applied to downscale original precipitation data from the GCMs. Taylor diagram and interannual variability skill methods were used to evaluate performance of the downscaled GCM data for simulating the EPIs. The results displayed that: (1) The downscaled 27 GCMs performed well for the wet EPIs. Based on the multi-model ensemble results, more severe and frequent precipitation extremes over 2015–2100 were projected. (2) The largest changes in extreme precipitation intensities over 2015–2100 would occur in central and southeast China. The maximum changes in extreme precipitation frequencies would occur in Tibetan Plateau and central and southern China. (3) Precipitation extremes in Xinjiang and Tibetan Plateau would be more sensitive to climate change than other sub-regions. The future EPI generally have longer than 10-year of primary periods. (4) The GCMs had larger contribution than the SSPs to the uncertainty in projecting the 10 EPIs. (5) CMIP6 models performed better than CMIP5 models in reproducing interannual variations of EPIs. This study supplies useful information to cope with future intensified precipitation extremes in China.