Observations of shear banding in fragmenting cylinders and in dynamic punch tests were analyzed for guidance in development of a theoretical model. The fragmenting cylinder tests led to a micro-mechanical model containing nucleation and growth processes for micro shear bands. The model constructed from the cylinder observations was able to predict the critical impact velocity for punching through a steel plate and to correctly represent the extent and nature of the damage in the plate for a range of punch velocities. Shear band damage to the punching rod was not well described by this model. Instead, it was necessary to expand the stress-strain relation to include thermal softening and to delay the onset of band nucleation in the model to approximate the band locations and orientations in the rod.