All civil engineering structures are deteriorating in time, at a different rate. The current and future states of a structure are associated with various degrees of uncertainty, and, therefore, a probabilistic approach is appropriate in order to predict future needs for maintenance and repair. Some existing models use discrete-time Markov processes called Markov chains. In this paper, homogeneous Markov processes (HMPs) with a finite set of condition states and a continuous parameter (time) are used. The aim is to make a probabilistic forecast of the condition of the structure, using time as a continuous parameter, so that probabilities of condition states can be calculated at each moment. The model is applied to the calculation of probability of possible future deterioration of the piers of Krk bridge (Croatia) that have already been repaired.