The combustion or pyrolysis of perfluorinated resins, under cer tain laboratory conditions (e.g., the "NIST" smoke toxicity test), produces fumes which are 2-3 orders of magnitude more toxic than the smoke from common plastics or wood. To determine whether these materials show the same high level of toxicity under realistic fire conditions, a series of twenty full-scale test burns was carried out at the facilities of the Laboratory for Heat Transfer and Fuel Technology, State University of Ghent, Belgium. A metal tray holding up to 30 kg of 25-pair telecommunications cable in sulated and jacketed with fluorinated materials (Teflon® FEP and/or Teflon® PFA) was exposed to fires involving 110 kg wood cribs, or energetically equiva lent amounts of diesel fuel or polyurethane foam. The burn facility was an 8' × 12' × 8' high (2.4 m × 3.7 m × 2.4 m) masonry room connected to a 43- foot (13 m) corridor. Smoke flowed out of the burn room, down the corridor, and into a smoke collection stack. Near the end of the corridor, a portion of the smoke was extracted, diluted and cooled, and passed through one of a series of stainless steel exposure chambers, each containing ten male Sprague-Dawley rats. Temperature and fuel weight loss were measured at regular time inter vals in the burn room, as were smoke temperature and composition (CO, CO2 and fluorine analysis), as it passed down the hallway and into the smoke collec tion stack. Carbon monoxide, oxygen, fluorine content and temperature of the smoke were also monitored in the animal exposure chambers. Fires were allowed to burn for 30-40 minutes. Animals were exposed to smoke for 30 minutes, after which the surviving animals were observed for a two-week post-exposure period. Selected animals were killed and tested for blood carboxyhemoglobin (COHb) content and tissue samples from the respira tory tract were subsequently examined for histopathological characteristics. Smoke reaching the end of the corridor contained only about one-third of the theoretical amount of fluorine. Additional losses were incurred during dilution and in the animal chamber. The fluorine loss is probably attributable to deposi tion of hydrogen fluoride in the high-humidity conditions associated with com bustion. The exposed animals showed some of the effects previously attributed to combustion products of fluoropolymers, but also showed near-lethal amounts of blood COHb, attributable to carbon monoxide from the principal fuel. Simi lar results were obtained whether the fire fuel was wood or diesel fuel. Use of polyurethane foam as a fuel was found to be unsuitable. The lethal smoke con centration of the cable smoke alone, i.e., without the effects of the carbon monoxide contributed by the principal fuel, is estimated to be 1.6 mg/l, or about 80 times less toxic than would be expected based on the NIST test. This toxicity is within a factor to two of what would be expected if the princi pal toxic agent (in addition to CO) were hydrogen fluoride or carbonyl fluoride. No evidence was found for the highly toxic agent present in the NIST test, which either did not form or existed only transiently in these experiments.