Architecture of biological systems is important for their homeostatic functions and is predictive of pathology. Volumetric imaging of intrinsic density, anisotropy, and 3D orientation of cell and tissue components is informative, but still challenging. These physical properties can be described succinctly by the volumetric distribution of the specimen’s permittivity tensor (PT). We report uniaxial permittivity tensor imaging (uPTI), a novel approach for label-free volumetric imaging with diffraction-limited resolution. uPTI combines the oblique illumination and the polarization imaging with high numerical apertures to encode the specimen’s permittivity tensor into intensity modulations, which are decoded with a novel vector diffraction model and a multi-channel convex optimization. The uPTI volumes of polystyrene beads and laser-fabricated anisotropic glass targets show that 3D uPT measurements are quantitative, with diffraction-limited spatial resolution of 0.23 × 0.23 × 0.8 μm3. Automated 2D and 3D imaging of a mouse brain section reveals anatomy at multiple scales (cm – 250 nm) and demonstrates that uPTI enables analysis of the density, anisotropy, and orientation of axon bundles and single axons. We multiplex uPTI with fluorescence de-convolution microscopy to enable correlative analysis of physical and molecular architecture of iPSC-derived cardiomyocytes. uPTI can be added to an existing widefield microscope as a low cost module. We share our implementation of the forward model and optimization algorithms via open source repository. Collectively, the reported advances in optical design, image formation, reconstruction algorithms, and biological interpretation open multiple avenues to study architecture of organelles, cells and tissue sections, including human cells and tissues that are challenging to label.