The steadily growing traffic load has resulted in lots of bridge collapse events over the past decades, especially for short-to-medium span bridges. This study investigated probabilistic and dynamic traffic load effects on short-to-medium span bridges using practical heavy traffic data in China. Mathematical formulations for traffic-bridge coupled vibration and probabilistic extrapolation were derived. A framework for extrapolating probabilistic and dynamic traffic load effect was presented to conduct an efficient and accurate extrapolation. An equivalent dynamic wheel load model was demonstrated to be feasible for short-to-medium span bridges. Numerical studies of two types of simply-supported bridges were conducted based on site-specific traffic monitoring data. Numerical results show that the simulated samples and fitting lines follow a curve line in the Gumbel distribution coordinate system. It can be assumed that dynamic traffic load effects follow Gaussian distribution and the extreme value follows Gumbel distribution. The equivalent probabilistic amplification factor is smaller than the individual dynamic amplification factor, which might be due to the variability of individual samples. Eurocode 1 is the most conservative specification on vehicle load models, followed by the BS5400 specification. The D60-2015 specification in China and ASSHTO specification provide lower conservative traffic load models.