This dissertation focuses on developing a predictive method for determining the dynamic densification behavior of thermite powder mixtures consisting of equivolumetric mixtures of Ta + Fe₂O₃ and Ta + Bi₂O₃. Of primary importance to these highly reactive powder mixtures is the ability to characterize the stress at which full compaction occurs, the crush strength, which can significantly influence the stress required to initiate reaction during dynamic or impact loading. Examined specifically are the quasi-static and dynamic compaction responses of these mixtures. Experimentally obtained compaction responses in the quasi-static regime are analyzed using available compaction models, and an analysis technique is developed that allows for a correct measurement of the apparent yield strength of the powder mixtures. The correctly determined apparent yield strength is combined with an equation of state to yield a prediction of the shock densification response, including the dynamic crush strength of the thermite powder mixtures. The validated approach is also extended to the Al + Fe₂O₃ thermite system. It is found that accurate predictions of the crush strength can be obtained through determination of the apparent yield strength of the powder mixture when incorporated into the equation of state. It is observed that the predictive ability in the incomplete compaction region is configurationally dependent for highly heterogeneous thermite powder systems, which is in turn influenced by particle morphology and differences in intrinsic properties of constituents (density, strength, etc.).