Road safety is a broad topic. It includes many aspects such as road conditions, vehicles, the environment, and drivers. Understanding how every factor contributes to the frequency of accident is not a simple task. Modelling road traffic accidents frequencies at hazardous road sections provides comprehensive insights into the safety effects of explanatory variables contributing to occurrence of accidents, and thus a variety of modelling techniques were used depending on the variable of interest. Since the number of Collisions consists of a nonnegative integer, count data modeling is commonly used in accident frequency analysis. In this study, both the Negative Binomial model and the Poison regression model were estimated. The empirical results are of great interest. The presence of national highways, posted speed limits, presence of drainage system, and the presence of crash Barrier increase frequency of accident at hazardous road sections. Nevertheless, the presence of right alignment, roads with 2 ways lanes, the presence of vertical and horizontal signs and road without parking. The research findings present advantageous insights on hazardous road locations in Tunisia and present useful planning tools for public authorities in Tunisia.