The development and calibration of both a plane-headed (hemi-spherical view), and a spherical-type heat flux meter has been undertaken in this work. These instruments have been shown to be capable of producing a signal which is directly proportional to the incident radiant heat flux. Both radiant heat flux meters developed in this work, make use of a radial disc, conductivity type sensing element, where the incident radiant energy is distributed radially via the disc to a cooled metal block. The metal block heat sink is located at the end of a water-cooled arm to enable insertion into high temperature environments. Transient response analysis of the plane-headed heat flux meter yields a time constant of 10. 6 seconds. A perturbation analysis of the spherical heat flux meter concluded that the response time is a function of the radiation heat transfer coefficient existing between the probe and its environment. A finite difference analysis has been carried out on the radial disc assembly in order to investigate the temperature distribution under steady state conditions. It has been concluded that the mode of attachment of the radial disc assembly onto the cooling water probe, can have a modifying effect on the magnitude of the heat meter signal. However, this effect does not introduce non-linearity into the steady state signal response. For the finite difference analysis, an empirical correlation has been derived describing the convective heat transfer at a plane surface with the flow of cooling water perpendicular to the surface. The correlation applies for annular flow, and is given as: Nu = 1. 045 Re0. 4 Pr1/3 Testing of the spherical heat flow meter has been carried out in a 440kW gas-fired furnace. It has been concluded from these trials that: (i) a peak signal output is obtained for an equivalence ratio, o of between 1. 15 and 1. 32, in the range of firing rates 118kW to 142kW, where o is defined as. (ii) the ceramic shield, which forms an integral part of the heat meter, did not develop cracks or physical defects during the trials, (iii) the peak signal from the heat meter closely follows the peak heat gain by the furnace cooling water load rather than the optimum combustion conditions, as indicated by the flue gas composition. A steady state mathematical model of the gas-fired furnace is presented here, and is compared with the results obtained from the furnace runs. This is the first stage in the development of an unsteady state furnace model for use as an aid in the testing of furnace control systems.