Fault tree analysis (FTA) is a logically structured process that can help identify potential causes of system failure before the failures actually occur. However, FTA often suffers from a lack of enough probabilistic basic events to check the consistency of the logic relationship among all events through linkage with gates. Sometimes, even logic relationship among all events is difficult to determine, and failures in system operation may have been experienced rarely or not at all. In order to address the limitations, this paper proposes a novel incident tree methodology that characterizes the information flow in a system instead of logical relationship, and the amount of information of a fuzzy incident instead of probability of an event. From probability statistics to fuzzy information quantities of basic incidents and accident, we pro- pose an incident tree model and incident tree analysis (ITA) method for identification of uncertain, ran- dom, complex, possible and variable characteristic of accident occurrence in quantified risk assessment. In our research, a much detailed example for demonstrating how to create an incident tree model has been conducted by an in-depth analysis of traffic accident causation. The case study of vehicle-leav- ing-roadway accident with ITA illustrates that the proposed methodology may not only capture the essential information transformations of accident that occur in system operation, but also determine the various combinations of hardware faults, software failures and human errors that could result in the occurrence of specified undesired incident at the system level even accident.