We describe a joint inversion approach that combines geophysical and thermal-hydrological data for the estimation of (1) thermal-hydrological parameters (such as permeability, porosity, thermal conductivity, and parameters of the capillary pressure and relative permeability functions) that are necessary for predicting the flow of fluids and heat in fractured porous media, and (2) parameters of the petrophysical function that relates water saturation, porosity and temperature to the dielectric constant. The approach incorporates the coupled simulation of nonisothermal multiphase fluid flow and ground-penetrating radar (GPR) travel times within an optimization framework. We discuss application of the approach to a large-scale in situ heater test which was conducted at Yucca Mountain, Nevada, to better understand the coupled thermal, hydrological, mechanical, and chemical processes that may occur in the fractured rock mass around a geologic repository for high-level radioactive waste. We provide a description of the time-lapse geophysical data (i.e., cross-borehole ground-penetrating radar) and thermal-hydrological data (i.e., temperature and water content data) collected before and during the four-year heating phase of the test, and analyze the sensitivity of the most relevant thermal-hydrological and petrophysical parameters to the available data. To demonstrate feasibility of the approach, and as a first step toward comprehensive inversion of the heater test data, we apply the approach to estimate one parameter, the permeability of the rock matrix.