Railway level crossings are critical elements in railway and road networks with accident occurrences resulting in fatal and severe injuries. In addition to the human loss, level crossing accidents also negatively impact rail transport reliability and transport speed. For safety management, specific risk factors should be identified and their impact on overall safety quantified. To this end, multivariate regression equations, commonly known as accident prediction models, have been used in the study. The paper describes the development of accident prediction models in three Central European countries (Czech Republic, Hungary and Austria), using samples of data on railway level crossings with flashing lights. The models were used to quantify the impact of several risk factors. The cross-country study design enabled comparison of obtained experience and drawing conclusions for further development of both road and railway network safety management.