In this study, dual angle observations of radiative surface temperature have been used in conjunction with a two-layer model to derive sensible heat flux over a sparsely vegetated surface. Data collected during the semi-arid-land-surface-atmosphere program (SALSA) over a semi-arid grassland in Mexico were used to assess the performance of the approach. The results showed that this approach led to reasonable estimates of the observed fluxes. The mean average percentage difference (MAPD) between observed and simulated fluxes was about 23%, which is not statistically different from the expected 20% scatter, when different flux measuring devices are compared over the same site. However, the sensitivity analysis indicated that the approach was rather sensitive to uncertainties in both measured radiative temperatures and aerodynamic characteristics of the vegetation. Finally, the issue of using dual angle observations of surface temperature for characterizing the difference between aerodynamic and nadir viewing radiative temperature has been examined. The results showed that this difference is linearly correlated with the difference between nadir and oblique radiative temperatures. Based on this finding, we expressed sensible heat flux in terms of the (nadir) radiative-air temperature gradient and a corrective term involving the nadir-oblique temperature differences. This formulation has been successfully tested. The resulting MAPD was about 33%. © 2001 Elsevier Science B.V.