The aim of the present work is to extend the adaptive multigrid Local Defect Correction (LDC) [1] method to nonlinear solid mechanics problems. Adaptive local multigrid approaches are efficient adaptive mesh refinement methods that have been initially introduced for computational fluids dynamics problems. They have only been recently successfully applied to solid mechanics [2, 3]. Studies conducted in the linear elasticity context have revealed the efficiency of the LDC approach compared to the widely-used h-adaptive methods [2, 4]. The main advantage of the LDC method lies on the partitioning of degrees of freedom between levels of refinement, which may be still beneficial while dealing with nonlinear problems. In this work, a special attention is devoted to methodological and algorithmic modifications required to adapt the LDC method to the nonlinear solid mechanics framework. Especially, prolongation and restriction operators applied to link levels of grids are pointed out. We also discuss the best way to carry out the two embedded iterative processes: nonlinear problem’s resolution and LDC multigrid resolution. Moreover, fundamental features related to the projection and conservation of fields as well as internal variables between levels of refinement over time are addressed. Performances of the LDC method are tested on benchmark problems. In order to automatically drive the mesh adaptation, the a posteriori recovery-based Zienkiewicz and Zhu error estimator as well as a mesh optimality criterion aiming to globally and locally satisfy a given accuracy are used.