Although this material has been generally regarded as prone to exothermic self-decomposition with very rapid release of heat, oxygen and gaseous chlorine derivatives it has not been the subject of an orthodox thermal ignition study such as that performed on anhydrous calcium hypochlorite (UN 1748), studied in the standard manner in 1978. This paper describes such tests on samples varying in mass from 5 g up to 200 kg in cylindrical containers of varying aspect ratio and various materials including stainless steel gauze, high-density polyethylene and coated fibre. The heat transfer coefficients of all the containers were measured accurately by both steady-state and cooling curve procedures and the thermal conductivity of the hypochlorite was measured independently. These measurements revealed that the Biot numbers of the samples tested were in a range where the critical ambient temperatures were sensitive to the values and thus sensitive to the convective airflow around the test bodies. With these factors taken into account the usual Frank-Kamenetskii (F-K) plot gave a good straight line in the range 115°C upwards. However, below this temperature, i.e. in the range 40-115°C very strong deviations occur. Nevertheless, the F-K theory still holds as the low-temperature points also fall on a good straight line with activation energy 48.5 kJ/mol which compares with the high-temperature activation energy much greater than this. These results indicate a sharp change in the rate determining step for heat generation at temperatures around 100-120°C. In temperature ranges both above and below this area the temperature-time traces for the samples behave classically as would be expected from the F-K theory but inside this range the traces show much more complex behaviour with some oscillatory characteristics, also some evidence of endothermic behaviour. The consequences of this unexpected behaviour for safety (particularly in the bulk marine shipping context) are serious. The deviations predicted from simple extrapolation of the high-temperature results indicate much lower critical ambient temperatures for large quantities of this material than previously thought. Our low-temperature results obtained by orthodox ignition methods are entirely consistent with independent direct isothermal calorimetric measurements already present in the literature but not used quantitatively in this context.