Accidents do not occur by chance. An accident results from the complicated interaction of various causes. Identification and prioritization of hazardous road sections can not be achieved without determining effects of these causes on accident occurrence. In this paper, having effective factors in accident occurrence known, a model was proposed for identifying and prioritizing hazardous road sections. This model was developed based on nature of accident and using multi-attribute decision-making methods. One of the most important results of this research is that outputs from prioritizing hazardous road sections based on accident cause may be more accurate and different from those of current models which don't take accident cause into account. According to the developed model, a segment may be compared to another and be identified to be high accident-prone with the same cause, whereas, the total number of accidents occurred in the former may be less than that in the latter. The model was evaluated based on the real data collected from highway segments in Tehran.