Transport systems are more and more complex and the number of users is increasing. So, it makes necessary both to increase the availability of the material and to guaranty an high level of security keeping reasonable maintenance costs. For this reason, the maintenance optimization of transport systems has become a key issue. Currently, industrial jointly use systematic maintenance and corrective maintenance. For the reasons mentioned above, to optimize the logistic behind the maintenance, they want to anticipate the failure time. That’s why they recently interested to the predictive maintenance based on prognostic concept. Dynamic Bayesian networks (DBN) are powerful mathematical tools that can model behaviour of complex systems. The Markov property in a DBN imply that the sojourn time in each degradation state is exponentially distributed. To overcome this limitation, Duration variables have been used. The Diagnostic and Maintenance group from IFSTTAR-GRETTIA developed a model called VirMaLab (Virtual Maintenance Laboratory) based on DBN and GDM to describe the degradation of a system and his diagnosis/maintenance process). This approach was already successfully applied to evaluate and optimize maintenance strategies of several railway systems, from various rails application to some rolling stock components. If Bayesian networks were already used in some prognosis application, their contribution was generally limited to provide some static decision making, not to estimate directly the RUL. The originality of the proposed study lays in the integration of a dynamic RUL estimation represented with an explicit node in a DBN. In this paper, an online prognosis algorithm and its representation in a dynamic Bayesian network are presented. This algorithm is a first step to the development of a online predictive maintenance policy that could be optimized using its representation in the VirMaLab DBN.