Coal permeability is sensitive to the effective stress and is therefore coupled to the geomechanical behaviour of the seam during gas migration. The geomechanical response of the seam is complicated by the additional effect of coal shrinkage with gas drainage. Existing coal permeability models, such as those proposed by Shi and Durucan (2004) and Palmer and Mansoori (1996), simplify the geomechanical processes by assuming uniaxial strain and constant vertical stress. Connell (2009) and Connell and Detournay (2009) investigated these assumptions using a fully coupled simulator. An important obstacle to the use of the coal permeability models of Shi-Durucan or Palmer-Mansoori is the characterisation of the various physical properties involved, since a number of difficult to measure properties are involved. This paper summarises the influences on coal permeability and in particular the role of coal shrinkage with gas desorption and compressibility with pore pressure and the importance these processes could have on predictions of gas production. The paper considers the assumptions involved in the derivation of the models to describe coal permeability and the impact these could have on predicting gas production. A novel model is derived which is more general and this serves as a basis for the discussion presented in this paper. Measurements of coal properties (cleat compressibility, geomechanical properties, Biot coefficient) with respect to pressure are presented for two Australian coals (a sample from the Hunter Valley and one from the Bowen basin) and the measured behaviour compared with that assumed in the Shi-Durucan and Palmer-Mansoori coal permeability models derivation.