1 INTRODUCTION During their service life, structures that are subjected to an aggressive environment may suffer degradation of their component materials. Such damage could decrease the mechanical charac- teristics of the system, and, with them, even its reliability. In this paper, special attention is paid to the durability of concrete structures subjected to the diffusive attack from environmental ag- gressive agents, like sulfate and chloride, which may lead to deterioration of concrete and corro- sion of reinforcement (Biondini et al. 2004a). The great randomness connected with the occurrence of critical environmental attacks sug- gests studying deterioration of materials and structures from a probabilistic point of view (Gara- vaglia et al. 2002, Biondini et al. 2004b). To this purpose, in this paper the deterioration process over time will be defined as a stochastic process of the random variable (r.v.) τ which represents the “lifetime” of the system spent in a given service state. In this way, the deterioration process is viewed as a dynamic process where the system loses its performance during time (loss of re- sistance, loss of reliability, change of quality,....). In particular, when the loss reaches a certain level, the system suffers a “damage” and it passes from the present service state i to another service state k characterized by a lower level of performance. Each changing of state happens after a time τ ik , defined as the waiting time of the system in the state i before to pass in another state k (Garavaglia et al. 1997). Under this point of view, the deterioration process can be con- sidered as a transition process and it can be modeled as a semi-Markov Process (s-MP) with one step of memory (Guagenti et al. 1988). Based on such criteria, the present study is aimed to develop a probabilistic model able to predict the service life of deteriorating structural systems for which the deterioration process can be modeled as a s-MP. The application to a concrete box-girder cross-section shows that the proposed methodology is effective even when the prediction is based on a few observed data. 2 RELIABILITY OF MATERIALS SUBJECTED TO A CRITICAL ENVIRONMENT The deterioration of the building materials subjected to environmental attacks and, as a conse- quence, their durability, may be dealt as a reliability problem where (Sarja, 1996): Markovian Modeling of the Time-Variant Deterioration Process of Concrete Structures Fabio Biondini & Elsa Garavaglia Department of Structural Engineering, Politecnico di Milano, Milan, Italy Franco Bontempi Department of Structural and Geotechnical Engineering, University of Rome “La Sapienza”, Rome, Italy ABSTRACT: This paper presents a probabilistic model aimed to predict the service life of dete- riorating structural systems. In the proposed approach, damage evolution is modeled as a semi- Markov process with one step of memory. Specific attention is paid to the use of the model in the assessment of the time-variant probabilistic performance of concrete structures in aggressive environments. The application to a concrete box-girder cross-section shows that the proposed methodology is effective even when the prediction is based on a few observed data. IABMAS’04, Second Int. Conf. on Bridge Maintenance, Safety and Management, Kyoto, Japan, October 18-22, 2004