Using the CN05.2 dataset of daily mean temperature for the period from 1961 to 2014, total extreme cold events (TECEs) are determined in the present study. These TECEs are classified into two categories according to the domain they affected in eastern China (East of 105°E in China), i.e., the first category of extreme cold events (FCECEs) that only affected Northeast China and the second category of extreme cold events (SCECEs) includes those that affected most of eastern China. Results show that the trend of intensity of FCECEs is significantly negative, but the opposite is true for that of SCECEs on the interdecadal timescale. As for the occurrence frequency, the trend for SCECEs is obviously negative, and that for FCECEs is also negative during the pre-1990s epoch but becomes positive after the 1990s. The occurrence frequency of TECEs is the highest in February among the winter months, and it tends to decrease during the pre-1990s epoch but increases after the 1990s. Northwesterly winds (cold air surges from Lake Baikal) at 850 hPa during SCECEs is stronger than that during FCECEs, and the opposite is true for northeasterly winds (cold air surges from the Okhotsk Sea). Analysis of the E-P flux and divergence field at 300 hPa indicates that significant wave-like anomalies appear across Northeast China during both SCECEs and FCECEs, and the Rossby wave energy propagation is weaker during SCECEs than during FCECEs. Harmonic analysis shows that the overlap of longwave trough and ultra-longwave geopotential height anomaly in the same phase can more easily lead to extreme cold event.