A total of 28 foam/fabric upholstery mock-up combinations (7 foams and 4 fabrics) were tested in a rate of heat release calorimeter (the cone calorimeter) at incident fluxes of 17.5 and 25 kW/m 2 . The results were analysed with three objectives: (a) understanding the relative contributions of fabric and foam; (b) comparing the relevance of the incident flux used to fire hazard; and (c) comparing the results with those of a round robin carried out on another rate of heat release calorimeter: Ohio State University instrument.The fabrics used represented the commercial range, but had no particular additional degree of flame retardance added: a cotton, a nylon, a polyester and a wool-nylon fabric The foams used were all flexible polyurethane foams, but had very different densities and levels and types of additives. All foams contained at least some flame retardant additives.The results suggest conclusions, at least for the systems tested:1. The fabric plays a much greater role than the foam in determining the fire performance of the system. Downloaded from 224 2. Ignition of the system is essential to obtain an adequate reflection of rate of heat release and fire performance. 3. If the incident flux is too low, results can be misleading, because some fabrics will not ignite.4. Trends on heat release are the same for the cone calorimeter and the OSU calorimeter, when the incident flux is high enough for the systems to ignite. 5. These systems do not reflect the highest degree of fire performance attainable ; however, the largest opportunities for improvement in upholstery systems will be found by using fabrics with better fire performance or by using interliners, rather than by addressing the foams alone.