The land surface skin temperature diurnal cycle (LSTD) is an important element of the climate system. This variable, however, cannot be directly obtained globally from polar orbiting satellites because such satellites only pass a given area twice per day and because their infrared channels cannot observe the surface when the sky is cloudy. To obtain the skin temperature diurnal cycle and fully utilize satellite measurements, we have designed an efficient algorithm that combines model results with satellite and surface-based observations and interpolates satellite twice-daily observations into the diurnal cycle. Climatological information from a climate model, CCM3/BATS, is used to determine a normalized shape (typical pattern) of the diurnal temperature for different latitudes, seasons, and vegetation types. The satellite observations, which are by themselves inadequate, are combined with the normalized modeled diurnal typical patterns to obtain the skin temperature diurnal cycle. The normalized typical patterns depend on the parameters of the diurnal insolation, such as sunrise, sunset, and peak times, and are also affected by the type of vegetation cover and soil moisture. The underlying physical foundation of this algorithm is that the diurnal cycle of temperature can be viewed as a composite of a daily average, diurnal periodic component, and random aperiodic component (noise). With the assumption that the noise can be ignored, the daily average can be inferred from satellite twice-per-day measurements and the periodic part can be obtained from modeled climatologies, providing a reasonable approach for estimation of the diurnal cycle of skin temperature. The general framework of the algorithm and its application for the clear-sky (cloud free) conditions are presented. This cloud-free version of algorithm is evaluated using FIFE and BOREAS field experiment surface observations. Regional tests over the Mississippi River basin have also been conducted using GOES-8 and AVHRR observations. Uncertainties of this cloud-free algorithm have been analyzed, indicating an accuracy of about 1--2 K for monthly cloud-free diurnal cycles at near-pixel resolution. ¿ 1999 American Geophysical Union