Literature up to 1975 is reviewed and it is concluded that although the mechanism of heat transfer in fire-test furnaces is well understood, there is no foundation on which to base a rational procedure for calibration of fire-test furnaces. A procedure is proposed for indicating differences between furnaces by relating real gas temperature, measured temperature and radiation temperature. The method was applied to three SP furnaces using suction pyrometers and radiation meters. Significant differences between furnaces were found, mainly related to furnace size. Experimental results agreed satisfactorily with theoretical calculations. Examples are given for insulated steel structures and wood charring calculations, showing that furnace characteristics can affect fire test results.