Membrane penetration in high pressure triaxial tests were measured by two different methods and some discussions were made on the results obtained. For the dense sand, it was found that the membrane penetration obtained by the two methods gave approximately the same value. For the loose sand, however, the method based on the assumption that the material was isotropic, gave a significantly larger amount of membrane penetration than that measured by the alternative extrapolation method in which a set of compression tests were performed on several specimens which contain metal rods of different diameters. This was considered to be caused by the anisotropic fabric of the loose sand. Typical relation curves between membrane penetration and hydrostatic pressure were presented to be used for the correction of membrane misfit in high pressure triaxial tests.