Abstarct: baxsed on traffic accident statistics, it is clear that traffic accidents impose the harmful effects and unnecessary costs on society; thus, researchers try to analysis traffic accidents, identify accident locations, and determine significant parameters.Today, data mining techniques, specifically spatial clustering methodes, are used to analyze accident data and find their spatial patterns. In this research, a combined method called GriDBSCAN algorithm was used for clustering accident data. The DBSCAN algorithm applied for its spatial characteristics and the GRID algorithm used by taking data into a grid mesh increased the accuracy and execution time in big data such as accident data. Other clustering methods such as K-Means, Nnh, KDE and spatial autocorrelation were also used to be compareed with GriDBSCAN method. The results showed that Nnh algorithm was the most accurate method for spatial clustering of traffic accident points, and GriDBSCAN algorithm was also applicable to the separation of high density areas of accidents in high volume data. The clusters obtained from GriDBSCAN algorithm properly differentiated the accidents of different urban areas according to the density. By examining the parameters in these clusters, it is possible to reduce the number of accidents and understand the factors affecting them.