The Markov chain (MC) model has been widely used for predicting deterioration of bridges around the world. The main component of the MC model is a transition probability matrix, which can be used to determine the probability of a bridge being in each of a set of condition states in the future based on its current condition state. Accurate definition of this matrix is therefore vital for ensuring good performance of the MC model. The maximum-likelihood estimation method (MLE) along with actual inspection data can be employed to determine this transition matrix. This study first goes over the MLE method for the determination of the Markov transition matrix. Following that, Markov transition matrices for six different US states, namely, Texas, Georgia, Mississippi, New York, Ohio, and Wisconsin were determined using the MLE method and bridge inspection data. In the end, the times to reach each condition state were determined for the concrete bridge decks in each location. This study shows that the concrete deck service life in the northern states with cold climates is significantly lower than that in the three states in the south with warm climates.