Three-dimensional, matrix-based statistical analysis methods were developed and integrated with high-resolution topographical imaging, to assess how microstructural changes influence the evolution of plastic deformation and strain localization in a commercial AA5754-O aluminum sheet in three in-plane strain modes. Analysis of the raw surface data revealed that the general composition of the surface roughness was highly sensitive to strain mode and strain level. The microstructural conditions that promote strain localization were assessed by extending a profile-based surface roughness parameter (Rt) to matrix form. Both analyses revealed that different strain modes produce characteristic dissimilarities in the deformation at the grain level. The localization data can be well characterized with a two-parameter Weibull distribution, suggesting that strain localization is a stochastic process that can be modeled reliably with Weibull statistics. This study clearly demonstrates that an accurate and straightforward probabilistic expression that captures the microstructural subtleties produced by plastic deformation can be developed from rigorous analyses of raw topographic data. Because variations in surface morphology profoundly influence the reliability of the numerical models used to predict strain localization, incorporating expressions of this type could greatly enhance the accuracy of these models.