An understanding of the relative permeability of gas and water in coal reservoirs is vital for coalbed methane (CBM) development. In this work, a prediction model for gas-water effective permeability is established to describe the permeability variation within coal reservoirs during production. The effective stress and matrix shrinkage effects are taken into account by introducing the Palmer and Mansoori (PM) absolute permeability model. The endpoint relative permeability is calibrated through experimentation instead of through the conventional Corey relative permeability model, which is traditionally employed for the simulation of petroleum reservoirs. In this framework, the absolute permeability model and the relative permeability model are comprehensively coupled under the same reservoir pressure and water saturation conditions through the material balance equation. Using the Qinshui Basin as an example, the differences between the actual curve that is measured with the steady-state method and the simulation curve are compared. The model indicates that the effective permeability is expressed as a function of reservoir pressure and that the curve shape is controlled by the production data. The results illustrate that the PM-Corey dynamic prediction model can accurately reflect the positive and negative effects of coal reservoirs. In particular, the model predicts the matrix shrinkage effect, which is important because it can improve the effective permeability of gas production and render the process more economically feasible. (C) 2013 Elsevier B.V. All rights reserved.