Terahertz (THz) spectroscopy can potentially be used to probe and characterize inhomogeneous materials, however spectroscopic identification of such materials from spectral features of diffuse returns is a relatively underdeveloped area of study. In this thesis, diffuse THz scattering from granular media is modeled by applying radiative transfer (RT) theory for the first time in THz sensing. Both classical RT theory and dense media radiative transfer (DMRT) theory based on the quasi-crystalline approximation (QCA) are used to calculate diffuse scattered intensity. The numerical solutions of the vector radiative transfer equations (VRTE) were coded and calculated in MATLAB. The diffuse scattered field from compressed Polyethylene (PE) pellets containing steel spheres was measured in both transmission and reflection using a THz time domain spectroscopy (THz-TDS) system. Measurement results showed energy redistribution by granular media due to volume scattering as well as angle dependent spectral features due to Mie scattering. The RT model was validated by successfully reproducing qualitative features observed in experimental results. Diffuse intensity from granular media containing Teflon, lactose sugar, and C4 explosive was then calculated using the RT models. Simulation results showed the amplitude of diffuse intensity is affected by factors such as grain size, fractional volume of grains, thickness of scattering layer, and scattering angles. Spectral features were also observed in the diffuse intensity spectra from media containing grains with THz spectral signatures. The simulation results suggest the possibility of identifying materials from diffuse intensity spectra.