We investigate the decrease in permeability, k, with depth, z, in the Oregon Cascades employing four different methods. Each method provides insight into the average permeability applicable to a different depth scale. Spring discharge models are used to infer shallow (z 0.8 km, our results agree with the power law relationship, k = 10-14 m2 (z/1 km)-3.2, suggested by Manning and Ingebritsen <1999> for continental crust in general. However, for shallower depths (typically z ≤ 0.8 km and up to z ≤ 2) we propose an exponential relationship, k = 5 ¿ 10-13 m2 exp (-z/0.25 km), that both fits data better (at least for the Cascades and seemingly for continental crust in general) and allows for a finite near-surface permeability and no singularity at zero depth. In addition, the suggested functions yield a smooth transition at z = 0.8 km, where their permeabilities and their gradients are similar. Permeabilities inferred from the hydroseismicity model at Mount Hood are about one order of magnitude larger than expected from the above power law. However, higher permeabilities in this region may be consistent with advective heat transfer along active faults, causing observed hot springs. Our simulations suggest groundwater recharge rates of 0.5 ≤ uR ≤ 1 m/yr and a mean background heat flow of Hb ≈ 0.080--0.134 W/m2 for the investigated region.