A finite volume technique in a commercial computational fluid dynamics (CFD) code is employed in this study to simulate transient, incompressible, non-Newtonian and non-isothermal rubber mixing. The simulation processes are conducted in a two-dimensional(2D) domain, where a mixing chamber partially-filled with rubber is equipped with a pair of two-wing non-intermeshing counter-rotating rotors. The main objective is to assess the effect of different fill factors of rubber on dispersive and distributive mixing characteristics by simulating 15 revolutions of the rotors rotating at 20 min −1 . 50%, 60%, 70%, 75%, 80% and 90% are the six different fill factors chosen for the study. An Eulerian multiphase method has been applied to solve for the two different phases, rubber and air. The non-Newtonian property of rubber is handled using the shear rate dependent Carreau-Yasuda model, along with an Arrhenius function to include the temperature dependency. In addition to the governing equations related to the conservation of mass, momentum and energy, the volume of fluid (VOF) method is chosen to track the interface between air and rubber. With regard to the results, flow patterns, thermal distributions, viscosity behavior and volume fraction are analyzed for the different fill factors. In addition, dispersive and distributive mixing behavior is also assessed in detail using Lagrangian statistics, such as mixing index, cumulative distribution of maximum shear stress, cluster distribution index (CDI), scale of segregation (SOS) and length of stretch (LOS), calculated from massless particles. Both the Eulerian and Lagrangian results showed that fill factors between 70% and 80% presented the most reasonable and efficient mixing scenario, and also exhibited the best dispersive and distributive mixing characteristics combined.