Abstract <p>Zirconium alloys are the predominant cladding materials in current nuclear reactors, and their creep behavior under high temperature plays a decisive role in structural safety. Meanwhile, it is crucial to obtain accurate creep predictions based on a small amount of data due to the limited data and high testing costs of creep experiments. In the study, the axial creep behavior of Zr alloy tubes was tested on a Zwick universal testing platform at 350°C under different loads. To improve both the physical consistency and predictive accuracy of the model, a sinh-type constitutive equation was adopted, and model parameter identification together with uncertainty quantification was carried out using Bayesian inference and the Markov Chain Monte Carlo (MCMC) algorithm. This framework yields fully probabilistic predictions, including lifetime credible intervals. The results demonstrate that the model not only accurately captures the experimental data, but also exhibits strong extrapolation capability and reliability in the low-stress regime. Thus, this study achieves material life prediction through constitutive modeling and probabilistic inference framework.</p>