Understanding of the detailed mechanisms of how roots anchor in and reinforce soil is complicated by the variability and complexity of both materials. This study controlled material stiffness and architecture of root analogues, by using rubber and wood, and also employed real willow root segments, to investigate the effect on pullout resistance in wet and air-dry sand. The architecture of model roots included either no laterals (tap-root) or a single pair at two different locations (herringbone and dichotomous). During pullout tests, data on load and displacement were recorded. These studies were combined with Particle Image Velocimetry (PIV) image analysis of the model root-soil system at a transparent interface during pullout to increase understanding of mechanical interactions along the root. Model rubber roots with small stiffness had increasing pullout resistance as the branching and the depth of the lateral roots increased. Similarly, with the stiff wooden root models, the models with lateral roots embedded deeper showed greatest resistance. PIV showed that rubber model roots mobilized their interface shear strength progressively whilst rigid roots mobilized it equally and more rapidly over the whole root length. Soil water suction increased the pullout resistance of the roots by increasing the effective stress and soil strength. Separate pullout tests conducted on willow root samples embedded in sand showed similar behaviour to the rigid model roots. These tests also demonstrated the effect of the root curvature and rough interface on the maximum pullout resistance.