The undrained behaviour of sands is affected by many factors including the initial state and loading path. If laboratory studies are to be meaningful in the context of in-situ problems, both the initial state and subsequent loading path in-situ have to be appropriately simulated in the laboratory. This is critical since most natural soils have an anisotropic (or technically a transversely isotropic) fabric and their response is direction dependent. The term initial state encompasses the density, fabric of the soil, and the stress conditions which are characterized by the three principal stresses and their directions. This paper summarizes the impact of anisotropic fabric on both monotonic and cyclic undrained behaviour of sands. The anisotropic response that is manifested in simpler triaxial compression and extension tests and is systematically studied using a relatively complex hollow cylinder torsional shear device. It is shown that the tendency for strain softening, which could lead to catastrophic consequences under monotonic loading, is very much dependent of fabric and loading path. It is demonstrated that current cyclic liquefaction design practice that depends on the cyclic resistance ratio, CRR is rather simplistic, but the available data suggests that cyclic simple shear resistance is possibly a lower bound for the actual resistance. The data and experience presented in this paper suggest that the difficulties encountered in using laboratory test results in field problems are possibly a result of inappropriate laboratory characterization without regard to the initial fabric and the subsequent loading paths in-situ.