A new method has been introduced to predict impact sensitivity of different types of polynitroheteroarenes which can include nitropyridine, nitroimidazole, nitropyrazole, nitrofurazane, nitrooxadiazole, nitro-1,2,4-triazole, nitro-1,23-trazole and nitropyrimidine explosives. The model is based on the number of carbons and hydrogens as well as specific structural parameters that can increase or decrease impact sensitivity. The new method has been applied to 67 polynitroheteroarenes which have different structural parameters. The predicted results are compared with outputs of recent calculated results of complex neural network. The root mean squares (rms) of deviations of different polynitroheteroarenes are 36 and 66 cm for the new and neural network methods, respectively. The novel model also predicts good results for some miscellaneous nitroheterocyclic explosives and several new synthesized polynitroheteroarenes compared to experimental data. (C) 2008 Elsevier B.V. All rights reserved.