Road traffic safety has always been a major social problem in China. In order to get a better understanding of the types and distributions of road traffic accidents, this paper constructs the Accident Crash Scenarios(ACS) classification system based on the traffic accident data collected by the traffic management department in some place of China from 2013 to 2015, which select four influence variables on the basis of Crash Relevant Events(CRE) from Naturalistic Driving Data. The proportion of each variables are analyzed, and all ACSs are divided into 48 scenarios. The first nine ADSs of highest proportion are extracted from all 10596 ACSs, and Multi-factor Logistic Regression(MLR) and ANOVA analysis are used to research the ADS of the highest proportion involved the type of vehicles and pedalcyclists, and pedalcyclists include bicyclists, motorcyclists, tricyclists and electric bicyclists. Of all influence variables obtained from the classification system, two are found most significantly associated with the ACS involved pedalcylists, which namely weather and road type. The data from Naturalistic Driving Studies(NDS) in Shanghai are extracted then, the first seven CREs of highest proportion is also extracted from all 430 CREs based on 37 Pre-crash topology defined by NHTSA, and the comparison between ACSs and CREs confirms the risk scenario involved pedalcyclists is relatively dangerous, the possibility of causing ACSs is relatively high. Comparison on influence variables of risk scenarios involved vehicles and pedalcyclists in ACSs and CREs are relatively few, and the risk scenarios are indeed special typical scenarios in China. Research on this paper may contribute significantly for the improvement of road traffic safety in China.