This project evaluates the feasibility of re-identifying commercial trucks with data automatically collected at Oregon weigh-in-motion (WIM) stations. The methods consist of two main stages. The first uses a Bayesian model to match each vehicle from the downstream station to the most ÒsimilarÓ upstream vehicle. In the second stage, methods are introduced to screen out vehicles crossing only downstream and to adjust accuracy vs. total number of vehicles matched by calculating the highest and second highest similarity measures for each vehicle. The proposed approach improves the accuracy of re-identification significantly. The models are applied to data from three WIM stations, creating two different ÒlinksÓ of 125 and 145 miles respectively. The algorithms match around 95% of trucks crossing both sites with about 90% accuracy. A threshold parameter allows adjustment of accuracy vs. total matched vehicles. When travel times vary widely between sites the mismatch rate increases.