AbstractThe confinement effect of concrete by fiber-reinforced polymer (FRP) jackets may significantly enhance the axial compressive performance of concrete, although the results are variable. Therefore, rational calibration of deterministic models for compressive strength and the associated strain and stress–strain (SS) curves is required based on available experimental databases and selected probabilistic models. To improve the accuracy of the probabilistic model, second branch classifications for the stress–strain curve of FRP-confined concrete are identified within a series combination of uncertainties with computerized classification algorithms before probabilistic calibrations. The Bayesian theorem and Markov chain Monte Carlo (MCMC) approaches were used to update a probabilistic model that includes the critical variables that have been established in previous research. Furthermore, eight representative deterministic strength enhancement models, three strain enhancement models, and four stress–strain models were chosen for evaluation using credible intervals (CIs) and confidence levels (CLs) at different strain levels. Different types of FRPs were also analyzed individually to ensure the validity and reliability of the findings. The suggested probabilistic models can predict the properties of ultimate axial stress and related strain, thus offering an effective method for calibrating the confidence level and computational correctness of deterministic models previously published in the literature.