This paper proposed a novel mechanical model of nail joint restoring force with seven model parameters were based on low-cycle and repeated tests of light wood structure nail joints. This model considers different pinch points and slip effect in both positive and negative directions. Model parameters were identified by using the Bayesian method.The results show that the accuracy of Bayesian identification method relies on the quality of raw data. The result of the most probable value and the associated covariance matrix can be determined and used to the calculation of nail joint with the same conditions. The model response obtained from the result of parameters identification is in good agreement with the experimental results. The restoring force model proposed in this paper provides a reference for future nail joints analysis. Furthermore，Bayesian method is extended to parameter identification of nail joints restoring force，providing a novel way to investigate nonlinear hysteretic analysis of light wood structures.