SUMMARYEight free burning and two sprinklered fire tests were performed with electrical cable trays and live digital switch racks in a large enclosure to simulate telecommunications central office (TCO) fires started by electrical overheating. Very-slow-growing (non-flaming), slower-growing (partially flaming) and lowintensity-faster-growing (flaming) fires releasing gray-white, gray, and black smoke, respectively, were observed in the tests. Under quiescent conditions present in the unvented enclosure fire tests for cables, very-slow-growing fires were detected in about 1452 s, whereas the slower-growing fires were detected in about 222 s by commercial fire detectors. Under ventilation conditions typical of TCOs, detection times were very similar for the five types of commercial TCOs fire detectors used in the tests. The average detection times for slower-growing fires (cable fires) and low-intensity-faster-growing fires (digital switch rack fires) were 242 AE 17% and 249 AE 11%s respectively.The TCO procedures to reduce smoke damage from fires (on fire detection, inlet ventilation flow is turned off and exhaust flow is turned on) were found to be beneficial. The extent of smoke damage decreased significantly with an increase in the exhaust flow rate. The chloride ion mass deposition suggested that equipment recovery would be possible in the smoke environment if the cable vapor concentration could be reduced below about 3 g/m 3 . The metal corrosion rate was found proportional to the 0.6th power of the smoke concentration, similar to that found for the corrosion of metal surfaces exposed to aqueous solutions of HCl and HNO 3 and for acid rain with no protective layer at the surface. Sprinkler water was found to wash down the smoke deposits on the surfaces with little indication of corrosion enhancement.