Measurement of soil thermal conductivity in-situ is important for many practical geotechnical and civil engineering applications. However, for a variety of reasons it can be difficult to measure in-situ. A novel method to estimate the thermal conductivity of sands based on electrical relative permittivity (dielectric constant) measurements and GPR data is presented. First, an analytical model to estimate thermal conductivity of soils was developed and validated using a database of thermal conductivity measurements. The new thermal conductivity model was then compared to existing models using a larger database of new and pervious thermal conductivity measurements conducted on a wide range of soils. Second, a new model to estimate relative permittivity of soil was developed and compared to existing empirical models based on a new database of relative permittivity measurements on soils. The new model is based on particle level geometry with an empirical fit. Last, a method of estimating thermal conductivity from relative permittivity measurements was developed based on the previous two parts and was then extended to relative permittivity estimated from ground penetrating radar (GPR) data. Thermal conductivity and relative permittivity measurements were conducted on prepared bench scale specimens of five soils of varying density and saturation. In total, 124 specimens were prepared for thermal conductivity and relative permittivity measurements. Thermal conductivity was measured using a thermal needle technique and relative permittivity was measured using a Dynamax TH2O probe. Based on the laboratory bench scale tests, a method of estimating thermal conductivity from the relative permittivity measurements was empirically developed. The method was then tested using GPR by preparing large box specimens and collecting data on sand specimens of known dimensions and physical properties (void ratio, saturation, dry density). The effectiveness of estimating thermal conductivity from GPR measurements is presented.