Toxic product yields from five commercial cables obtained from a steady state tube furnace method , the Purser furnace,) are compared with results from a large scale test, which uses the physical fire model in the proposed prEN50399-2-2 test, with the addition of effluent gas analysis, using Fourier transform infrared (FTIR), and for further comparison, a static tube furnace method (NF X 70-100). This work represents one of the first attempts to establish a relationship between bench-and large-scale toxic product yields for burning cables. This is difficult because the cables have been formulated for low flammability, and do not, therefore burn consistently. The tube furnace burns the cable completely, whereas the large scale test effluent is the result of a combination of flame spread and toxic product yields, both of which are fire scenario dependant. There is significant differentiation between cable types based on composition, and arising because only a portion of the cables burn in the large scale test, accompanied by possible decomposition of hydrate sheaths. The fire stage of the large scale test appears to have been replicated in an appropriate manner, given the correspondence of the CO 2 /CO ratios. The yields of CO 2 , CO, HCl and smoke show reasonable agreement, given the differences in extent of burning, and the accuracy of the mass loss data available for the large scale test. The yields and extent of burning have been combined to demonstrate the estimation of toxic hazard for a particular fire scenario based around the large scale test, which shows only marginal sensitivity to the differences in toxic product yield between the steady state tube furnace and the large scale test.