We have used a split Hopkinson pressure bar arrangement to investigate impact-induced reaction of the secondary explosives HMX, RDX and PETN in granular form. Sentencing of the experiments was performed by detecting reaction light emission, the spectral analysis of which can also provide information about the temperature of reaction. We measure the fraction of the mechanical energy that passes through the specimens that is absorbed in the run up to reaction, which we refer to as the efficiency factor, and for these experiments is of order 5–10%. We postulate that the efficiency factor is a function of the microstructure. The measured amounts of energy that were absorbed are comparable to those amounts required to bulk heat the samples to their melt points. A critical absorbed energy for reaction implies a minimum duration of loading for a given mechanical power and efficiency factor, and this idea is supported by the observation that the more intense the loading, the shorter the time to reaction. Additionally, we postulate a critical minimum mechanical power below which heat is redistributed faster than it can be accumulated. A minimum mechanical power threshold in turn dictates a minimum pressure threshold; but the idea of a stand-alone critical pressure is not experimentally supported.