Abstract: Accurate earthquake prediction is one of the ways to effectively reduce earthquake disasters. Until now, empirical earthquake prediction methods have remained in use, but the accuracy falls far short of what is needed to effectively mitigate earthquake disasters. The trend in earthquake prediction lies in the transition from empirical prediction to physical prediction, and the key to this transition is to achieve a deeper understanding, depiction, and tracking of the seismogenic process. According to the experimental results of rock mechanics, it can be found that before failure, the rock mainly experienced a stress-increase stage and a subsequent stage of sub-instability. We have analyzed the seismogenic process of some natural earthquakes. In the analysis, apparent stress and b-value are used together to obtain information on the stress variability of the crustal medium. The correlation between seismicity and Earth rotation (described by p-value) is used to obtain information about the critical state or sub-instability phase of the crustal medium. Thus, the seismogenic process can be depicted. In this paper, based on a case study of the seismigenic process for some natural strong earthquakes, we propose an idea and method for earthquake prediction based on the tracking of the seismogenic process, and provide a basis for the identification of short- to mid-term siemic risk regions. It is expected to be beneficial for practical earthquake prediction.