In this paper, we present results of an integrated simulation study to assess the effectiveness of seismic imaging for monitoring CO2 sequestration. We considered two scenarios. In the first, injected CO2 remained confined within a shallow coal formation. In the second, the sequestered CO2 gas leaked through a semi-permeable shale layer to an overlying sand unit sealed above. In both scenarios, the CO2 injection process resulted in enhanced production of coal-bed methane. The reservoir and seismic simulations required the construction of 3D geologic and facies models, the estimations of seismic velocities based on rock physics correlations, and the development of geostatistical dual-porosity reservoir descriptions of the coal and overlying shale and sand units. We ran the 3D reservoir simulations with GEM 2008.10, an equation-of-state compositional reservoir simulator from Computer Modeling Group with the capability to model adsorption of injected CO2 in coal, desorption of CH4, and matrix shrinkage-swelling effects. The reservoir description was derived from a 3D geostatistical model of the Big George coal in the Powder River Basin. We used Batzle-Wang, Gassmann, and Mavko-Dvorkin-Mukerji relationships to incorporate the effects of mineral contents and changing reservoir properties on P-wave velocity. We built synthetic seismograms by simulating the propagation of acoustic waves through the 3D heterogeneous media, and used reverse time migration to create 3D seismic images corresponding to the state of the reservoir at the beginning and end of ten years of CO2 injection. The resulting seismic images clearly identified the regions of CO2 gas saturation, closely matching the gas-saturation profiles predicted by the reservoir simulator. Further research will be necessary to fine-tune an integrated reservoir simulation, rock physics modeling, and seismic imaging workflow to provide highly advanced imaging, monitoring, and verification capabilities for CO2 sequestration. This study is a first step toward this goal.