The probabilistic model using Markov Chain Monte Carlo (MCMC) sampling procedure was proposed for estimating the value of strut efficiency factor in the strut‐and‐tie model (STM). The STMs according to various proposals from different researchers were compared with the database containing 691 tests on the shear strength of RC deep beams. Strut efficiency factor models were collected and compared against the experimental results in the database. Considering prior information, established database, and existing models, the probabilistic model was improved by considering the shear span‐to‐depth ratio, web reinforcement ratio, longitudinal reinforcement ratio, and dimension of specimen. The proposed model was evaluated by the results in the database, which shows better accuracy and discreteness with mean and CoV of 1.047 and 0.248, respectively.