This report reviews the modelling of post-flashover fires and compares the various methods of predicting temperature versus time in post-flashover compartment fires, including the historical development of theoretical approaches. The report specifically addresses the use of the COMPF2 model as implemented in the COMPF2PC computer programme, as a prediction tool for post-flashover fire temperatures. Aspects of the computer code are compared with theory and experimental data. The results of many COMPF2PC simulations are compared with test fire data, in order to determine how best to characterise the input data to achieve the best simulation results with the computer programme. It is found that with careful selection of input data, COMPF2PC can provide good prediction of post flashover fire temperatures for compartments with a fire load of greater than15 kg of wood per square metre of floor area, and for ventilation factors A˯√H/A˕ ≥ 0.04. Reliability of temperature prediction is poorer for ventilation factors (A˯√H/A˕) significantly less than 0.04. Guidelines for use of the COMPF2PC programme are provided. Based on the methodology developed during simulation of test fires, generalised fire temperature versus time curves are developed for a single compartment size and a range of compartment material properties. The generalised COMPF2PC temperature versus time curves are compared with those of alternative models in common use. It is found that for a fire of fire load 1200 MJ m-² of floor area, in a compartment of medium thermal inertia, depending on ventilation, the COMPF2PC model predicts fires which either have a significantly higher maximum temperature or longer duration (or both), than those predicted by the Eurocode Parametric fire, and the "Swedish" fire model of Magnusson and Thelandersson. This may have a significant impact on the calculation of time equivalent fires. Recommendations for future development of the COMPF2PC programme are provided.