Fabric and composition of a series of Upper Permian high-volatile to low-volatile bituminous coals of the Sydney basin have a marked effect on stress sensitivity of permeability, and thus on the reservoir characteristics of the coal. The coals vary in composition from end members of predominantly bright-banded coal comprised mainly of the microlithotype vitrite and the maceral vitrinite, to dull coal composed of significant amounts of ash, inertinite group macerals, and the microlithotype inertite. The brighter coals are more extensively fractured with one or, more commonly, two or three regularly spaced fracture sets (cleats) spaced at 5-20 micrometers. Fusinite and semifusinite, common macerals in the dull coals, are characterized by phyteral porosity (mainly cell lumens) and fabric- selective intergranular porosity. The permeability of tested samples varies significantly with composition and effective stress. The fabric of the samples is the most important factor in determining permeability and stress sensitivity of permeability. Coals with the highest permeability are those with at least one well-developed, throughgoing fracture set; these samples generally include abundant vitrite bands. The lowest permeability samples are nonbanded, with an attrital fabric and significant authigenic mineralization. At 0.8 MPa effective stress, the permeability of the sample suite ranges from 10 to over 2X106 microdarcies, a difference of five orders of magnitude. Even at high effective stresses (i.e., 12 MPa), the difference in permeability between samples varies as much as three orders of magnitude. The variation in permeability with effective stress (stress sensitivity) is marked; a twofold increase in effective stress causes a sixfold decrease in permeability. At effective stresses of over 9 MPa, the permeability of all samples is less than 10 microdarcies. Samples with the greatest stress sensitivity (the greatest reduction in permeability with increase in effective stress) are either thinly and discontinuously banded, or massive and notably mineralized with either quartz or carbonate. Such results suggest that, at least for low-permeability coals, the matrix compressibility is equally or more important in determining stress sensitivity than fracture closure. In-situ permeability tests of four coals (whole bench) are greater by about one order of magnitude than the laboratory-tested samples at the same effective stress. Such results reflect the scale effect and the laboratory bias toward testing stronger samples that are less permeable. The calculated stress sensitivity of the in-situ tested coals compares closely to that measured in the laboratory. Because of the marked variation in permeability with coal fabric and composition, stratigraphic and lateral heterogeneity of coal are important considerations during coalbed methane exploration and exploitation.