Font Size: 

Predicting motor vehicle crashes has established in the last decades a new area of research that fulfills the automotive domain and transportation facilities. The mathematical instruments support engineers to detect different effects before, during and after crash scenarios that will enhance knowledge about safety performance of road traffic. Using the Romanian road accident database, from 2008 to 2012, a series of analyses was conducted, to evaluate the causes that produced fatal road accidents when passing through curved and inclined plane road segments. The generalized linear Poisson model (Poisson-GLM) is commonly used in modeling motor vehicle crashes and this method is the mathematical tool used in this research. Results show that most of the fatalities occurred on curved road segments and less on inclined plane. The following factors were found to be significantly associated with an increased risk of fatal road accidents: the absence of road shoulders, or the type of the shoulders, driving without license, unsafe changing lanes and weather conditions. Moreover, most of the fatal crashes were found to occur in urban areas and lateral collision of a vehicle to pole, or similar objects was found significant to the increase of the number of fatal crashes. These results show the need for safety measures on curved road segments and less on inclined plane, especially in urban areas where the traffic density is high, with focus on lateral collisions.