The research on the correlation structure of the stock market is an important basis for the investment portfolio establishment and risk management. Random matrix theory is an effective method to analyze the correlation structure. Through the analysis of the correlation structure of China SSE A Share before, during and after the stock market crash in 2015, we get the following conclusions. Empirical data indicating that there is a non-random structure in the market. The correlation structure of China’s stock market is time-varying. During the period of the stock market crash, the level of the deviation of the correlation structure is more significant than that before and after the crash. In the partial correlation matrix excluding the market index, there are some stability characteristics related to the stock market value.