The automobile insurance is an important branch of non-life insurance in many countries. In some of them, coming first in the composition of the total income of the insurance branch (in Algeria for example). The implementation of the Bonus-Malus system is one of the solutions chosen by insurance companies to increase efficiency in the field of automobile insurance. The main purpose of this study is to use the Markov chain theories to calculate the relative premiums associated with each level of the Algerian Bonus-Malus system. These premiums will generally depend on the priori ratemaking. The first result of this research is that, the system of automobile a priori ratemaking used in Algeria is not optimal because of the difference found between the tariff variables used by the insurance companions (gender, use and power vehicle tax) and the tariff variables generated by applying the Poisson model using the GEE technique. In addition, in our work, two Bonus-Malus systems with 9 classes based on the Markov Chain theory have been calibrated, where only the penalty changes according to the type of Bonus-Malus system, and we concluded that the priori ratemaking influenced the scale obtained.