The measurement of the diffuse radiation incident on a horizontal surface, a priori a straightforward task, is fraught with difficulties. It is possible to measure the diffuse radiation by three different techniques: two of which measure it directly and the third indirectly. The most accurate is the indirect one, which is based upon the concurrent measurements of the horizontal global and the normal incidence beam radiation. The disadvantage of this being the relatively expensive tracking system required for measuring the latter. The diffuse radiation can be measured directly with a pyranometer outfitted with either an occulting disk or shadow ring, which prevent the beam radiation from impinging on the pyranometer sensor. The occulting disk can provide accurate measurements of the diffuse radiation but it requires a relatively expensive sun tracking system in the east-west axis. The shadow ring is a stationary device with regard to the east-west axis and blocks the beam radiation component by creating a permanent shadow on the pyranometer sensor. The major disadvantage of the shadow ring is that it also blocks that portion of the diffuse radiation obscured by the shadow ring. This introduces a measurement error that must be corrected to account for that portion of the sky obscured by the shadow band. In addition to this geometric correction factor there is a need to correct for anisotropic sky conditions. Four correction models have been applied to the data for Beer Sheva, Israel and the results have been evaluated both graphically and statistically. An attempt has been made to score the relative performance of the models under different sky conditions. (c) 2007 Elsevier Ltd. All rights reserved.