Knowledge of water content is important for water resources management and risk prevention. For areas covered with forest, the presence of decomposed or undecomposed forest litter poses a real problem for the monitoring of volumetric water content. Traditional methods such as time domain reflectometry (TDR) are difficult to apply, as the inconsistence of the porous material often leads to density modifications close to the probe wires. The aim of this work was to develop a TDR probe that can be used for loose materials while ensuring optimum working conditions in terms of ease of insertion into the porous medium, contact with the surrounding porous medium, and homogeneity and confinement of electrical energy storage. The performance of the new probe was tested on three porous media: a sandy soil and decomposed and undecomposed litter taken from four different sources of forest material. A classical TDR three-rod probe and a coaxial probe were used for comparison. Permittivity data were analyzed with parameters related to either mineral or organic media and two physically based models. The comparison of the results necessary for the evaluation of the probe performance was performed using the criterion of mean absolute error and the associated standard deviation, calculated on the difference between the water content values measured by the TDR probes and those determined by gravimetric observations. For the sandy soil, the new probe gave results comparable to those of the coaxial probe but more accurate than those obtained with the three-rod probe. For the forest litter samples, the new probe systematically gave the most accurate results. Once calibrated, the new probe allows measuring volumetric water content in forest litter within an error range of 0.02 m3/m3.