Abstract: The shear capacity of ultra-high-performance concrete (UHPC) beam is a key index for the design and safety assessment of UHPC structures, and it is of great significance to accurately predict the shear capacity. However, there is no unified calculation theory and design method so far, due to the complex shear mechanism. This study develops probabilistic models for the shear capacity of reinforced UHPC beams based on Bayesian theory. Firstly, a database that comprises 155 shear test results of UHPC beams was collected. The calculation formulas suggested by JGJ/T 465−2019, by AFGC−2013, by JSCE−2006 and, by SIA−2016 were selected as the Bayesian prior models, and their accuracies and applicability were evaluated and analyzed. Then, two types of the information, the sample information and the prior information, were synthesized by the Bayesian approach, and the deviation correction terms were added to update the prior models, so as to establish the posterior probability models for the shear capacity of reinforced UHPC beam. Finally, combining the test data used to modify the model and 28 sets of test data collected by the extended tests, the rationality and superiority of the probabilistic models proposed were verified simultaneously. The study results show that the Bayesian approach takes full advantage of the prior information of unknown parameters and can, at the same time, consider the uncertainty well. Compared with the prior models, the calculation results of the Bayesian probabilistic models are in a better agreement with the experimental data, and the randomness is significantly reduced, which can reflect the shear capacity of reinforced UHPC beam more reasonably and accurately. On this basis, the Monte Carlo Simulation (MCS) approach was employed to analyze the reliability of the shear capacity of UHPC beams, and the resistance partial safety factor was further determined upon the target reliability index.