Correlation analysis has been widely applied for process detection and alarm analysis. However, traditional correlation analysis methods have been found inefficient due to the influence of process time delays on analysis results. In this paper, we propose an alarm correlation analysis method based on block matching similarities. By converting alarm data sequences to time node sequences, the block matching similarity (BMS) method is able to reduce the calculation burden of the correlation analysis. Considering time delays between the alarm variables, the improved maximal correlation coefficient method is presented to reduce the effect of time delays and to improve the accuracy of the correlation analysis. A numerical case and the Tennessee Eastman process are employed to demonstrate the effectiveness and efficiency of the proposed method.