This paper questions the possibility that the measure of concrete permeability in laboratory depends in some Instances on the size of the sample, i e that intrinsic permeability measurements exhibit a size effect Experiments were carried out on specimens of different size (the surface Interested by the flow ranging from 2 x 10(3) to 2 x 10(5) mm(2)) of an ultra high strength concrete heated at 523 15K for 30 days A first campaign was made using a Cembureau permeameter, with silicon masks put on the sample's flat surfaces to reduce the areas where inlet or outlet pressures apply and otherwise standard settings and procedures A second one was made on concrete hollow cylinders of 265 mm height and 350 mm external diameter, fixed between steel plates so as to make a canister that was filled with pressurised nitrogen (from 253 to 415 kPa in different tests) and observed during pressure decay for 2-4 days Further Cembureau tests were made to correlate the two campaigns Mathematical models were put forward to process the information gathered from these experiments, dealing with bending of flow lines in the first case and with unsteady flow in the second, and numerical computations made to estimate permeability starting from boundary data measured in experiments A general analysis, based on the extreme value theory, supported the idea that a Frechet law should be adopted to possibly describe the growth of permeability with the size of the cross-section of the flow Though explorative in principle, our results lead to the conclusion that a size effect was observed The shape parameter of the Frechet law describing this effect was identified accordingly