A physical algorithm for calculating cloudy pixel skin temperature from its neighboring clear pixels is proposed generalizing that of Jin <2000>. Two neighboring pixels over the same land cover have a difference in temperature largely explained by surface insolation. Hence this algorithm starts from the surface energy balance equation (SEB) and expresses each energy term of SEB as a function of skin temperature. Then SEB is solved to derive cloudy pixel skin temperature from neighboring clear skin temperature plus a correction term determined by surface insolation, air temperature, and wind speed. This algorithm can be used for nights and winter hemisphere high latitudes where there is no surface insolation and is applicable to any surface where the principle of SEB is applicable. The algorithm is evaluated by using FIFE and BOREAS field experiments. Its global application has been examined through simulations with the NCAR Community Climate Model (CCM) coupled with the Biosphere-Atmosphere Transfer Scheme (BATS) simulations. The accuracy of the algorithm is encouraging: the rms error at a monthly pixel level is 1--2 K. Observed errors are greater when there is precipitation. ¿ 2000 American Geophysical Union