There is a lack of consensus on which microscopic geometrical properties are needed to predict macroscopic transport properties, many of which are direction-dependent, in natural and manufactured porous media. While the hydraulic conductivity K(q)is known to be a second-rank tensor, it is still commonly estimated from a scalar water retention y(q) function using statistical pore-size approach with a scale pore interaction term. We hypothesize that anisotropy in K(q) is caused by pore-scale heterogeneity resulting from the combined effects of grain fabric and lamination due to the alignment of aspherical soil particles. We suggest that K(q)can be predicted from y(q) using a tensorial pore interaction term. This hypothesis was tested by measuring directional hydraulic properties on a range of sediments using the Unsaturated Flow Apparatus (UFA). The traditional sample cell was modified to hold a cubic sediment sample which, starting at saturation, was centrifuged under constant speed for a predetermined number of step increases in speed to determine y(q) and K(q). Directional measurements were made on samples ranging from coarse sands to laminated silts and pure mica packed or collected in cylindrical and cubic sample holders. Results show a directional dependence in of both y(q) and K(q)with the difference between directions increasing as mean grain diameter decreased and aspect ratio of the particles increased. Measured K(q)was well described with commonly used functions based on statistical pore-size distributions by incorporating a tensorial pore interaction term.