Damage caused by incidents with transport tanks with compressed liquified gas is amongst the most extreme that can be encountered with transport vessels. This is particularly the case with the Boiling Liquid Expanding Vapor Explosion (BLEVE), which may occur if such a tank is exposed to fire for a prolonged period. Therefore, the local Dutch LPG transport sector adopted a thermally insulating tank coating as a 'standard outfit' for their tank trailers, with the aim to delay a BLEVE for a sufficiently long period for emergency services to take appropriate measures and for people near the accident location to be evacuated. On a European scale however, no consensus has been reached on the cost-benefit of such measures. With the current drive towards "greener" and renewable energy sources, this issue has regained attention with alternative fuels such as LNG, CNG and Hydrogen and a need was felt for (better) theoretical models and experimental data concerning the behavior of transport tanks carrying these substances. In this paper a new tank thermal (equilibrium) model is described to predict pressure and temperature behavior of a multi layered, thermally insulated tank containing a compressed liquified gas exposed to heat. Results are compared with data of three bonfire experiments, in which 3 m 3 tanks, filled for ca. 50% with LPG were exposed to fire. A good match between modelled and experimental pressure and temperature evolution in time could be obtained using a constant value for the thermal conductivity of the insulation layer. The modelling showed that the thermal insulation value is crucial for an accurate prediction of these parameters as well as the opening time for a pressure safety valve. As relevant temperatures may cover a very wide range (from cryogenic in LNG-tanks to over 1000 degrees C in a fire) knowledge of the thermal (and physical) behavior of the insulating layer over a large temperature range is essential. The same holds for the behavior of the PRV when subjected to fire. Extreme temperatures may also lead to deviating behavior from what is expected based on the initial settings.