Based on the daily maximum temperature data (CN05.1) in China during 1961-2017, this paper reveals the dominant modes of the interannual variation of the number of summer extremely high-temperature days (EHTD) in China through empirical orthogonal function analysis (EOF), and explores the key influencing factors and related physical mechanisms that lead to the formation of each mode. The results show that the first mode is the zonal distribution across China, which is mainly related to the Arctic Oscillation (AO). The Rossby wave train propagating southward from the northern Europe strengthens the zonal high anomaly across China when the AO is in a positive phase. The second mode shows the meridional dipole pattern, which is mainly influenced by the Polar-Eurasian teleconnection (POL) wave train spreading from North Atlantic Ocean to East Asia and the sea surface temperature anomaly over the western tropical Pacific enhances local Hadley cell, which makes the south (north) of China under the control of high (low) pressure systems. The increase of EHTD in the first two modes is related to the increase of incident solar radiation caused by the decrease of precipitation caused by local high-pressure anomalies. The distribution of the third mode is concentrated in the Tibetan Plateau (TP), which is mainly influenced by the zonal wave train propagating downstream from the Mediterranean Sea. On the one hand, the circulation anomaly corresponding to the wave train will cause the divergence of water vapor and the weakening of upward motion, which will decrease the cloud cover and increase the downward cloudy-sky shortwave radiation; on the other hand, it will cause atmospheric warming, which will increase the downward clear-sky longwave radiation. Both of them provide favorable conditions for the increase in the number of extremely high-temperature days. The results of this study will help to further understand the characteristics of summer extreme high temperature in China, and provide a theoretical basis for the seasonal prediction of extreme high temperature.