Abstract: Regional Persistent Extreme Precipitation Events (RPEPEs), characterized by wide-range, long-time and high-intensity precipitation, usually lead to severe disasters. Based on 7 RPEPEs occurred in Southwest China from 1999 to 2019, the sub-seasonal prediction performance of the FGOALS-f2 S2S model for RPEPEs in Southwest China was evaluated. The main results are as follows: FGOALS-f2 underestimated the summer precipitation in the Sichuan Basin and its surrounding areas, which resulted in the predicted extreme precipitation thresholds being about 15 mm/d lower than the observations in the Sichuan Basin, Chongqing, and most of Guizhou, while the thresholds were close to observations in the high-altitude areas of western Sichuan and Yunnan. The simulation error of precipitation intensity further led to the underestimation of the intensity of RPEPEs and short-term extreme precipitation events in those regions, but the errors did not change with the prediction leading time. Regarding the occurrence probability of extreme precipitation, the probability of detection, false alarm ratio and Heidke skill score of the RPEPEs and short-term extreme precipitation events decreased rapidly within the leading 1~10 days and then remained nearly constant during the leading 11~30 days. However, the prediction performance of RPEPEs was significantly better than that of short-term extreme precipitation. It can be seen that the prediction skill of the occurrence probability of the RPEPEs is of higher scores, which can provide scientific reference for the sub-seasonal prediction of extreme precipitation in Southwest China.