Three aging models, including Arrhenius,time-temperature superposition principle and neural network were used to comparatively study the aged HMX based PBX. And the lifetime was predicted when this explosives lost 0.1% of their total mass at 20 ℃ and when the lost 0.2% of the total mass at 50 ℃ and 60 ℃. Results show that the lifetime predicted from Arrhenius, time-temperature superposition principle, neural network are 390 years, 490 years and 15.2 years and 1127 d, 1180 d, 1375 d; and 182 d, 196 d, 220 d, respectively. It is stated that the results calculated by Arrhenius is agreed with that by Time-temperature superposition principle, and the neural network model is not an available when below the modeling temperature.