The first-order Markov Chain (MC) is used to predict the degradation of three types of pavements (rigid, semi-rigid, and mix) utilizing database in the five departments in the West of France. The assessment of uncertainty in the MC evolution is presented through studying the trend of mean and standard deviation, for components of the transition probabilities (TP) using different time steps (2, 3, 4, 5 and 6 years). The results show that the trend of rigid pavements is constant with time in terms of coefficient of variation. For semi-rigid and mix pavements, the trend of the standard deviation was constant with time. These statistical properties offer the opportunity to provide uncertainty modeling of TP. The propagation of uncertainty for 2 and 6 years time steps through the prediction of pavement condition index is also performed for analyzing the effect of the uncertainty. We compare the profile of states obtained from each time step in view to analyze the short (2 years) and medium term (6 years) potential of prediction.