Solids mixing affects thermal and concentration gradients in fluidized bed reactors and is, therefore, critical to their performance. Despite substantial effort over the past decades, understanding of solids mixing continues to be lacking because of technical limitations of diagnostics in large pilot and commercial-scale reactors. This study is focused on investigating mixing dynamics and their dependence on operating conditions using computational fluid dynamics simulations. Toward this end, fine-grid 3D simulations are conducted for the bubbling fluidization of three distinct Geldart B particles (1.15 mm LLDPE, 0.50 mm glass, and 0.29 mm alumina) at superficial gas velocities U/Umf=2-4 in a pilot-scale 50 cm diameter bed. The Two-Fluid Model (TFM) is employed to describe the solids motion efficiently while bubbles are detected and tracked using MS3DATA. Detailed statistics of the flow-field in and around bubbles are computed and used to describe bubble-induced solids micromixing: solids upflow driven in the nose and wake regions while downflow along the bubble walls. Further, within these regions, the hydrodynamics are dependent only on particle and bubble characteristics, and relatively independent of the global operating conditions. Based on this finding, a predictive mechanistic, analytical model is developed which integrates bubble-induced micromixing contributions over their size and spatial distributions to describe the gross solids circulation within the fluidized bed. Finally, it is shown that solids mixing is affected adversely in the presence of gas bypass, or throughflow, particularly in the fluidization of heavier particles. This is because of inefficient gas solids contacting as 30-50% of the superficial gas flow escapes with 2-3x shorter residence time through the bed. This is one of the first large-scale studies where both the gas (bubble) and solids motion, and their interaction, are investigated in detail and the developed framework is useful for predicting solids mixing in large-scale reactors as well as for analyzing mixing dynamics in complex reactive particulate systems. (c) 2017 American Institute of Chemical Engineers AIChE J, 63: 4316-4328, 2017