This paper presents the results of CO2 enhanced coal bed methane (ECBM) core floods on intact coal core from the Bowen Basin and the Hunter Valley, Australia, at pore pressures of 4 MPa and 10 MPa. The core floods involved flooding with CO2 to displace methane from the core and then reversing the flood by injecting methane to displace the CO2 from the previous flood. An important parameter for ECBM is the displacement or sweep efficiency which was estimated directly from the mass balance over the core flood. Displacement efficiencies obtained through CO2 injection were excellent with more than 99% of the CH4 recovered during the core floods. The reverse experiments in which CH4 was injected to displace CO2 were notably less effective with an average of 95% displacement obtained for the Bowen Basin core sample and only 71% displacement obtained for the Hunter Valley core sample by the end of the experiment. History matching was performed with the reservoir simulator SIMED II which used a hydrostatic permeability model, the extended Langmuir model, and a bi-disperse diffusion model. In general, good history matches were obtained between simulated and observed flow rates, mass balances, and breakthrough times demonstrating that the model could accurately represent the ECBM process. It was found that the triple porosity gas diffusion model provided an improved agreement to observations over the unipore model. Connell Lu Pan's hydrostatic permeability model was used in the history matching which differentiates between bulk and pore sorption strain. During the CO2 flooding experiments a change in permeability was observed as CO2 displaced CH4 in the core. As the stress conditions were constant, this was the result of the sorption strain impacting on the porosity and thus permeability. However, for the reverse core flood in which CH4 was injected to displace CO2, no permeability changes were observed, implying that pore and bulk strain were the same and thus cancelled out. (C) 2014 Elsevier B.V. All rights reserved.