Abstract We analyze the relationships between filter bandwidth, density of stations in the study area, threshold for correlation coefficient, and recognition rate. We also discuss the differences between “repeating events” by cross-correlation of seismic waveforms and the overlap of rupture zone semi-quantitatively. The results indicate that: more stations, local earthquake data, a 0.2-20 Hz bandpass filter and a detection threshold over 0.8 should be used. Furthermore, if MS≥4.5 for two repeating events, there is at least some overlap between their source zone.