A yaw sphere-thermometer assembly (YST), to measure sensible heat flux density by the eddy correlation method, was built following the design of Tanner and Thurtell (1970). Wind tunnel experiments indicate that the sphere constant should be 1.57, which is significantly less than the previously used theoretical value of 2.25. The effects of tilt indicate that heat fluxes may be in error by 5 per cent per degree of tilt in unstable, and up to 11 per cent per degree in stable conditions. A modified thermometer assembly was found necessary to provide durability. Field comparisons of the heat fluxes measured by the yaw sphere-thermometer system and a Bowen ratio apparatus produced satisfactory agreement. Direct measurements of sensible heat fluxes over a grass surface at Ladner, B.C. indicate a diurnal course very similar to that of the net radiation. In general, half-hour averaging periods showed no phase lag between sensible heat and net radiation. Field comparison of two YST systems gave good and consistent agreement. At a height of 2 m above ground and a horizontal crosswind separation of 1.5 m, less than 5 per cent variability was noted in the heat flux measurements from the two systems. For a 19 m horizontal separation, the variability was found to be less than 20 per cent. It is shown that the parameter (a), advanced by Priestley and Taylor (1972), can be a useful climatic indicator. The applicability of the eddy correlation technique to the measurement of sensible heat transfer between the atmosphere and the urban interface is demonstrated for a limited area of the city of Vancouver, B.C. Despite the enormous complexities of the turbulent heat exchange processes, the urban sensible heat flux pattern, obtained directly at heights of 1.2, 2, 4 and 20 m above roof-top level, largely reflected time and magnitude changes in the net radiation field, during the daytime. Nocturnal urban sensible heat fluxes, near roof-top level, were found to be directed away from the active surface. This is the reverse of the normal rural case. Within the local roof-top boundary layer, the sensible heat flux was found to be approximately constant with height and space (20% variation) during the daytime. At night, the existence of flux divergence and hence, non-constancy of the heat flux, is suggested. Daytime roof-top energy balances indicate that a significant portion of the net radiation is utilized in sensible heat transfer and in heat storage in the roof. The greatest energy is used in sensible heat transfer, which is about three times the heat storage at noon. With typical values of net radiation of 60 mWcm⁻², the sensible heat flux is about 30 mWcm⁻² and the heat storage 10 mWcm⁻². The residual term (equated to latent heat transfer) is quite appreciable. It is possible that the role of latent heat transfer is important for urban energy balance considerations. The nocturnal roof-top energy balance required a latent heat term of about 15 mWcm⁻² directed towards the active surface. The energy balance of the surrounding urban area was deduced from measurements of sensible heat flux and net radiation at heights above the roof-top boundary layer. On the assumption that these point measurements approximately reflect areally integrated averages, partitioning of the heat between sensible and latent heat yeilds a Bowen ratio of ~ 1 at midday.