A large number of bridges in Slovakia currently show signs of significant deterioration. These are predominantly prestressed and reinforced concrete bridges affected by chemicals used for winter infrastructure maintenance, mainly de-icing salts. Growing concerns about the progressive salinisation of concrete are creating pressure to refine the diffusion models currently used for predictions. Their reliability for structures in natural conditions proves problematic due to simplifying assumptions or the unknown inputs for the existing construction material. This article aimed to create a theoretical particle simulation model for a more accurate prediction of concrete contamination with chlorides promoted by carbonation based on data obtained from existing bridges. The proposed simulation model works based on Einstein’s – Brownian random walk particle model with consideration of non-interacting particle movement. The unknown diffusion properties of concrete are replaced by the probability of particle diffusion, which is obtained by iteration based on the satisfaction of convergence criteria determined from laboratory-measured concentrations of chlorides. The natural chloride profile is simulated considering a sharp-moving carbonation front, where different diffusion probabilities of particles are assumed in the carbonated and intact zone of the concrete at each time step. After obtaining the diffusion probabilities in the concrete cover, it is possible to predict the evolution of the chloride concentration. The paper further presents data obtained from two bridges whose piers are in close proximity to the traffic lanes.