Field permeability tests in cased boreholes have been modelled in a sand tank with controlled and measured hydraulic conditions. The testing program included "end of the casing" and "Lefranc" tests, as described in Canadian standards 2501-135 and 2501-130, for field permeability tests in a casing driven in a granular soil (valid for 10−5 < K < 10−2 cm/s approximately). The theoretical solutions for these tests require a list of assumptions that may be difficult to satisfy under field conditions. The method for detecting the common error in the assumed piezometric level gave in all cases a local piezometric level equal to that determined independently by a set of 22 piezometers. In accordance with the theoretical solutions, the K values given by the tests do not depend on the controlled upward or downward seepage in the sand tank. The average K value given by such field tests is strongly influenced by the preparation of the injection zone, which includes a natural soil in a more or less disturbed condition adjacent to an injection aperture or lantern of poorly known geometry and condition, so losses in hydraulic heads there may be different from those assumed in theory. From a practical point of view, all drilling operations have a major impact on the results and must be strictly controlled to have a reasonable estimate of the K value. Key words: permeability, field, borehole, model tests, piezometric level.