Factors contributing to traffic accidents are numerous, and they interact complexly. Even so, clarification of accident causes is indispensable in planning effective accident prevention measures. Onsite accident reports filled out by the police conventionally are used as the primary data for clarifying accident causes; however, a sufficient number of cases will never be obtained from such reports alone, no matter how long observations continue. For this reason, some studies use occurrence of traffic conflict as an alternative evaluation index. However, traffic conflict is rarely used in academic studies because of the difficulty in defining "complication" and the tendency toward subjective bias. This study proposes methods of investigating and analyzing data on the following: 1. accidents resulting in injury or death (based on onsite accident reports), 2. accidents involving property damage, and near-miss incidents (based on TAAMS images) and 3. traffic disturbances (based on videotaped images). It examines the validity of these analysis methods. Moreover, the influence of frozen road surface on traffic accidents in winter is examined by TAAMS images in this study.