The validity of the constant propagation fracture energy postulation for dynamic fracture is discussed. As shown from existing direct and indirect experimental results, thisassumption may not represent the prisical reality. For spontaneous fractures, the fracture energy was shown to increase linearly with the crack length, and for dynamic fractures driven by known amplitude impulsive loading (generated by planar impact), the fracture energy was not a constanteither. Despite of its phenomenogical origin, the Brobergs theory developed for self-similar crack growth works well for both spontaneous fractures and for dynamic fractures produced by well defined dynamic loading. In this theory, the fracture energy is not a constant. Furthermore, withgiven far-field loading or equivalent far-field loading, the crack speed is uniquely determined by a strength-like material parameter. This parameter is related to the cohesive strength proposed by H.J. Gao for perelastic materials in the crack-tip process zone. It is proposed in this work that the strength-like parameter (or equivalently the constant fracture speed) is a better material parameter to describe the dynamic fracture propagation process.