Cross-section deformation is considered an important indicator for assessing the structural safety in the inspection and maintenance of tunnels. The way it increases over its lifetime is an indication of the gradual degradation in structural performance. In order to take timely and appropriate maintenance measures before the tunnel reaches the ultimate limit state, a predictive degradation model of cross-section deformation should be established. In this paper, a probabilistic degradation model is developed based on an average uniform rigidity ring model for circular tunnels assembled from segmental linings. By considering the uncertainties and relevant performance of parameters that vary over time, the model is able to supply probabilistic and time-dependent predictions. Critical parameters are identified and the model is simplified following sensitivity analysis. Based on the measuring data, a Bayesian updating method is proposed to improve the input assumptions and predictions of the model. This research provides a perspective on the degradation modelling of the cross-section deformation of circular tunnels assembled from segmental linings and methods for improving the proposed predictive model.