Quantifying the magnitude and distribution of degree of saturation (Sr) in unsaturated soils is crucial to understand the grain-scale hydromechanical behavior, but it has been a major experimental challenge. This study proposes a new method to quantify the pore-water distribution non-uniformity, in terms of Sr, based on three-dimensional (3D) X-ray computed tomography images. The algorithm constructs vectors that consider the 3D spatial distribution of Sr for each REV. A weighted water distribution tensor (G, characterizing the spatial distribution of Sr) was derived to calculate a scalar parameter A that represents the non-uniformity. Application of the algorithm on sand samples demonstrated that the Sr distributions could be highly non-uniform at different drying states. The algorithm captured pore-water transport between the dilated and non-dilated zones of samples subjected to pre-peak shearing. The evolution of A with matric suction and axial strain showed potential in incorporating the pore-water distribution into microstructure-based constitutive models.