On the basis of simple physical arguments, the ratio of the weakest bond dissociation energy of nitro compounds to their decomposition enthalpy per covalent bond is put forward as a practical shock sensitivity index. Without any empirical fitting, it correlates remarkably well (R >= 0.95) with shock sensitivity data reported for 16 molecules spanning the most significant families of explosive compounds. This result supports the underlying assumption that this property depends on the ability of decomposition events to propagate into the material. It demonstrates that sensitivity structure relationships should take the energy content of the material into account. A linear regression against the present sensitivity index yields a predictive method with better performance than previous ones. Its sounder physical bases provide new insight into the molecular determinants of sensitivity and a compelling explanation for the sensitivity values reported for TATB and FOX-7.