Purpose: 4DCBCT reconstructs a temporal sequence of CBCT images, and often involves the binning of projection data to each temporal phase. Thus it suffers from binning errors during non-periodic breathing. The 5D model parameterizes respiratory motion by breathing amplitude and rate (e.g. tidal volume and airflow) independent of breathing irregularity. We propose to develop a new 4DCBCT reconstruction method based on the 5D model. Methods: The proposed CBCT technique requires that the breathing amplitude and rate are measured during CBCT. The proposed 5D-model-based method reconstructs a reference breathing phase, and time-independent deformation fields corresponding to the motion due to amplitude and rate. Therefore, the new method reconstructs 7 three-dimensional spatial maps, with reduced numbers of unknowns compared with the standard 4DCBCT method when 10 or more breathing phases are reconstructed. In addition, the 5D method explicitly manages non-periodic breathing.The 5D method is formulated as a simultaneous reference-image and deformation-field reconstruction problem, with total-variation (TV) regularization for both images and motions. Then the 4DCBCT reconstruction is carried out by alternately solving image reconstruction and image registration, in which the split Bregman method is used to reconstruct the reference image, and the Chambolle's duality-based algorithm is used to reconstruct the differential deformation fields. Moreover, the proximal terms are applied to ensure the convergence of the developed optimization algorithm for this nonconvex problem. Results: The proposed 5D method was validated in simulation based on measurements of a lung patient, in comparison with the state-of-art spatiotemporal-TV-based 4DCBCT iterative method. And the results show that the 5D method had significantly improved image quality and quantitative error, for both periodic and non-periodic breathing studies. Conclusion: We have developed a new 4DCBCT method based on 5D respiratory motion model, with improved image reconstruction during both periodic and non-periodic breathing studies. Jiulong Liu and Hao Gao were partially supported by the NSFC (#11405105), the 973 Program (#2015CB856000) and the Shanghai Pujiang Talent Program (#14PJ1404500).