Current and future detectors for high-energy particle physics like those at Large Hadron Collider (CERN) and EIC at Brookhaven National Lab pose high demands for the structural materials in a high-radiation environment where the accumulated radiation dose leads to defects like voids and cracks due to de-gassing and thermal cycling in the polymeric composite materials. In this work, terahertz time-domain spectroscopy (THz-TDS) is used for strain mapping of a polydimethylsiloxane (PDMS) doped with passive highly dielectrostictive strontium titanate (STO). A polarization polarization-dependent analytical model for the correlation of volumetric strain to the measured change in time of arrival for a THz pulse is developed. The model consists of effects due to changes in the dielectrostrictive properties of the composite due to changes in STO particle density and the change in thickness of the sample upon application of strain due to Poisson’s effects. The stress relaxation behavior of the composite is studied to avoid change in strain during the measurement window. The analytical model is validated with results using an open hole tensile and a circular edge notch specimen. The THz strain mapping results are compared with a scale-dependent finite element model (FEM) and surface strain measurement using the digital image correlation (DIC) method. The experimental results show sensitivity to material features like particle clumping and edge effects. THz strain map shows good agreement with FEM and DIC results proving the applicability of this technique for surface and sub-surface strain mapping in polymeric composites. The experimental results for THz-TDS-based volumetric strain maps agree with the results from DIC surface measurements as well as predictions using finite element models. This forms a robust analytical approach for the development of stress mapping and fracture front mapping in multilayer composites. The model inefficiencies at lower strain levels can be understood and mitigated by having better THz sensors and improving the signal-to-noise ratio.