[1] Air permeability is measured in the fractured crystalline rocks of the Roselend Natural Laboratory (France). Single-hole pneumatic injection tests as well as differential barometric pressure monitoring are conducted on scales ranging from 1 to 50 m, in both shallow and deep boreholes, as well as in an isolated 60 m 3 chamber at 55 m depth. The field experiments are interpreted using numerical simulations in equivalent homogeneous porous media with their real 3-D geometry in order to estimate pneumatic parameters. For pneumatic injection tests, steady-state data first allow to estimate air permeability. Then, pressure recovery after a pneumatic injection test allows to estimate the air-filled porosity.Comparison between the various studied cases clarifies the influence of the boundary conditions on the accuracy of the often used 1-D estimate of air permeability. It also shows that permeabilities correlate slightly with fracture density. In the chamber, a 1 order-ofmagnitude difference is found between the air permeabilities obtained from pneumatic injection tests and from differential barometric pressure monitoring. This discrepancy is interpreted as a scale effect resulting from the approximation of the heterogeneous fractured rock by a homogeneous numerical model. The difference between the rock volumes investigated by pneumatic injection tests and by differential barometric pressure monitoring may also play a role. No clear dependence of air permeability on saturation has been found so far.Citation: Guillon, S., M. T. Vu, E. Pili, and P. M. Adler (2013), Field and numerical determinations of pneumatic flow parameters of unsaturated fractured porous rocks on various scales, Water Resour.