Analysis of the torque data obtained for a large range of carbon blacks in an oil-extended butadiene rubber (CB-441) shows that the rate of decrease of torque (after the second power peak) follows first order kinetics. The rate of decrease represents the rate of reduction in effective filled volume fraction through dispersion of carbon black agglomerates, and thus, a reduction in the volume of rubber occluded between individual aggregates within the agglomerates. The assumption that the rate of torque reduction is proportional to the rate of carbon black dispersion was tested by examining the responses to various factors influencing the mixing process. In general, the conclusions reached from the analysis of torque data were in agreement with the common industrial experience and predictions based on the mathematical analysis of dispersive mixing. Tadmor's analysis of dispersive mixing predicts that the rate of agglomerate rupture depends on the number of particle-particle contacts and thus is related to the size of individual aggregates, but is independent of agglomerate size. Thus, it is in agreement with the present findings that the rate of dispersive mixing increases with decreasing surface area and increasing structure of aggregates. Increasing polymer-filler interaction gives rise to a faster rate of dispersive mixing, possibly by increasing the effective radii of aggregates through bound rubber formation. Increasing the batch temperature increases the rate of dispersive mixing due to reduced cohesion between the aggregates and a more favorable balance between cohesive and shearing forces. Increasing carbon black loading increases the rate of dispersive mixing by increasing the viscosity and, thus, shearing forces generated during the mixing process. The technique developed in this work may provide a better means for measuring dispersibility of carbon blacks, since other available methods suffer certain disadvantages. For instance, the resistivity measurements are not only dependent on carbon black dispersion, but also on the chemical nature of its surface, while microscopic methods depend on the examination of very small samples that may not be representative of the whole batch.