This paper describes some thermal analysis experiments conducted on high explosive samples. These employ differential scanning calorimetry to monitor thermal effects at elevated temperatures (around 200degreesC) and heat conduction calorimetry to record thermal effects at much lower temperatures (below 100degreesC). The work shows that, due to the generally high thermal stability of many high explosive compositions, heat generation rates are very low, if detectable at all, at normal storage temperatures, even when using a very sensitive instrument. The sensitivity and reproducibility of this technique has been investigated in detail by Wilker et al. [S. Wilker, U. Ticmanis, G. Pantel, Detailed investigation of sensitivity and reproducibility of heat flow calorimetry, in: Proceedings of the 11th Symposium on Chemical Problems Connected with the Stability of Explosives, Sweden, 1998] and shown to be capable of recording heat generation rates of less than a microwatt. This allows continuous measurement of decomposition processes in nitrate ester based propellants at temperatures as low as 40degreesC. However, the measurement of very low levels of heat generation is difficult, time consuming and therefore expensive. If the assumption is made that the life limiting process is invariably the slow decomposition of the energetic component, this will frequently lead to very long service lifetime predictions. A number of possible complications are identified. Firstly, due to its low detection threshold, a heat conduction calorimeter may detect other reactions which will not lead to failure, but which may still dominate the heat flow signal. Secondly, the true failure process may generate little energy and be overlooked. In view of these considerations, at present it seems unwise to rely on heat conduction microcalorimetry as the only tool for the assessment of the life of high explosive energetic systems. Based on examples of life terminating processes in high explosives during storage and use, it is clear that decomposition of the energetic material is not invariably the cause of system failure. It is also by no means the only reaction that may take place in, and be observed by, a heat conduction calorimeter. (C) 2003 Elsevier Science B.V. All rights reserved.