The present Diploma Thesis aims to investigate road accident severity per vehicle type. On that purpose, the database with disaggregate road accident data of the Department of Transportation Planning and Engineering of the National Technical University of Athens (NTUA) was exploited, and lognormal regression models were developed. The application of these models allowed the identification of the effect of various parameters such as the type of accident, the weather conditions, the lighting conditions etc. on the number of killed, seriously and slightly injured, per number of vehicles involved, for each vehicle type separately (car, truck, bus, motorcycle up to 49cc and motorcycle over 50cc). One of the results was that, for every vehicle type, accidents in which two or more vehicles are involved are more severe, due to the fact that typically they are higher-speed and hence higher impact. Furthermore, severity is higher at accidents involving pedestrians and passenger cars.