This Research Project evaluates the combustion severity of New Zealand upholstered furniture materials. Experimental combustion tests on typical upholstered furniture fabric and polyurethane foam combinations form the basis for all conclusions reached. 63 bench-scale Cone Calorimeter and 10 full-scale armchair Furniture Calorimeter combustion tests were conducted in the Fire Engineering Laboratory at the University of Canterbury. 7 different polyurethane foams, including 2 fire-retardant, are tested along with 100% polypropylene and 95% woollen fabrics. These tests demonstrate that the variation of foam and fabric covering play a substantial role in influencing the combustion characteristics. Between the wool and polypropylene fabric types, there were several combustion behavioural differences identified. Most significantly was the ability of the woollen fabric to remain in place under intense heat exposure for a longer time than the polypropylene. This had the effect of prolonging the ignition times in the Cone Calorimeter tests and increasing the time to peak heat release rates (HRRs) for both the Cone and Furniture Calorimeter tests. The effects of the various types of polyurethane foam were generally less significant than the effects caused by varying the fabric type. However, one type of fire retardant foam showed combustion characteristics that were significantly out of pattern from the others, by having prolonged ignition times and longer times to peak HRRs in the Cone and Furniture Calorimeter tests respectively. Thus the effects of the fire retardant foam was clearly shown to interfere with the combustion behaviour. All experimental methods in this Research Project follow the methods developed by the European fire research programme CBUF- Combustion Behaviour of Upholstered furniture. Thus, the results in this Research Project are meaningful on an international level. Model I, a method for predicting full-scale burning combustion characteristics from bench-scale test data, as developed by the European CBUF research, is applied to the New Zealand materials. The full-scale furniture combustion Model is compared in three areas, which are the value of peak HRR (kW), time to peak HRR (s) and the total amount of heat released (MJ), from burning full-scale armchairs. The Model does not accurately predict the full-scale burning characteristics, especially for the predicted time to peak HRR and total heat released. Instead the Model is conservative from a design perspective, predicting the time to peak HRR in a shorter time and a higher total heat release. For the peak HRR prediction, the Model achieves a level of confidence comparable with the European data that was used to validate the Model. Therefore it is considered accurate enough to be used to predict the peak HRR for the selected full-scale armchair style, without doing full-scale tests.