Within agroforestry systems, tree root architecture is a driver of important ecological processes such as nutrient cycling and carbon storage. Yet the belowground component of trees remains largely understudied due to methodological constraints. Conventional subsurface sampling can overlook the heterogeneity of root systems, while complete excavations are destructive and unrepeatable. Thus, there is a need to develop in situ non-intrusive technologies, such as ground penetrating radar (GPR), to measure root systems. Radar signals emitted from GPR into the subsurface are reflected at the coarse root-soil interface due to a contrast in dielectric properties. The returned radar signals can be interpreted to determine root biomass estimates and distribution. This technology was tested in two agroforestry systems: a tree-based intercropping (TBI) system in Canada and a cocoa-shade (CS) system in Ghana. Drawing on these tropical and temperate systems, we provide an overview of the methods used, the limitations of the technology, and the promising applications of GPR to study the ever-elusive tree root in order to answer important ecological questions pertinent to land management objectives. Using GPR image analysis, we successfully and non-destructively detected alley tree and shade tree plus cocoa tree-crop coarse root distributions in TBI and CS systems, respectively. We show strong root plasticity in agroforestry systems as: i) interspecies variability was detected in tree root stratification patterns and ii) coarse root distribution was modified in the presence of a secondary species and under contrasting edaphic conditions. In the TBI system, tree root biomass was accurately estimated for five tree species using GPR image processing when compared to the matched harvested root systems. This new application of GPR allows for the charting of tree root structure and a novel method for root biomass estimation, and therefore carbon storage, in agroforestry systems.