Distributive mixing in an internal mixer was studied numerically by means of tracking the evolution of the distance between pairs of particles in the mixing chamber. The distributions of these pairwise distances are reported in terms of the probability density function of a pairwise correlation function. In conjunction with this descriptive technique, a dynamic particle-tracking algorithm for two dimensions has been developed to study the dynamics of mixing in the mixer. We also propose a parameter epsilon to monitor the extent of mixing. This general approach represents the first attempt of this kind to address directly the goodness of mixing. A total of five operating modes was considered. Three of them have even speed ratio of 60 rpm and rotors positioned at 90-90, 90-180, 90-270 relative to the horizontal axis, respectively. The remaining two have uneven speed of 60 rpm/40 rpm and rotors positioned at 90-90, 90-270 relative to the horizontal axis, respectively. For each operating mode, a complete period is represented as a sequence of snap shots (72 for even speed and 216 for uneven speed) and the flow field for each snap shot was calculated by means of FIDAP. Based on our proposed framework, we were able to demonstrate clearly that the anti-symmetric configuration (90-270) is the best operating mode among the three even speed cases and the configuration (90-180) is the worst. The performance of the uneven speed cases fell in between the even speed cases.