The sections where risk of accidents is the highest in railway transport are the level crossings. In many countries, accidents that occur on the level crossings lead to loss of goods and lives. Therefore, governments have adapted some regulations to prevent accidents that occur at level crossings. In this study, we have examined the data of level crossings belonging to particular years in Turkey. The effects of physical and traffic characteristics of the level crossings, which are determined in the framework of the related regulation, are interpreted by logistic regression modeling method. As a result, it was determined "the width" and "the type" of level crossings are independent determinants of accidents, while "traffic moment" is not. Since the traffic moment parameter is not an independent determinant, the sub-parameters forming the traffic moment are used as inputs for the new model. In the new model, which sub-parameters of traffic moment, the type and width of the level crossing parameters are used as input, "the number of annual average traffic value of vehicles" and "the type of level crossing" parameters are found as independent determinants of accidents while the number of annual average traffic value of trains and the width are not.