Previous experimental and theoretical studies have examined the possible use of thermal-infrared measurements of surface temperature to estimate sensible heat flux from homogeneous surfaces. In extending the methodology suggested by Lagouarde and McAneney to larger scales, it is necessary to consider the effect of spatial variability. Over a heterogeneous mixture of surfaces, statistical errors are possible due to spatially-correlated changes in windspeed, surface roughness and surface temperature. Errors due to the latter two variables are investigated here at the 1 km2 scale of the NOAA AVHRR using two specific case studies. Errors due to correlated variations in windspeed and surface roughness, and windspeed and surface temperature, are also estimated. The results suggest that all these errors are generally negligible if application is restricted to surfaces having a maximum crop height of around 1 m. Given this qualification, estimates of regional scale sensible heat fluxes seem feasible using spatially-averaged measurements of surface temperature along with ancillary estimates of averaged maximum air temperatures, windspeed and surface roughness.