Material Point Method (MPM) simulations of large deformations in strain-softening materials often encounter strain localization, which prevents numerical solutions from converging to a single, physically consistent result upon spatial refinement. Existing strategies to address this pathological discretization sensitivity in MPM typically involve empirical modifications to material properties or require significant changes to the theoretical and/or numerical formulation. This work introduces a length-regularized MPM that embeds an intrinsic length scale into Bsplines weighting functions and employs them for interpolation within the MPM framework. By maintaining a fixed support size for the B-splines weighting function across different discretization levels, the method effectively circumvents the discretization sensitivity while preserving the governing equations and model formulation. Numerical simulations of strain localization due to damage and plasticity demonstrate the proposed method’s ability to robustly simulate large deformations while mitigating pathological discretization sensitivity.