Empirical correlations for predicting both beam and diffuse radiation at Beer Sheva, located in the semi-arid southern region of Israel, have been developed. These equations relate either the beam or the diffuse fraction of the global radiation to the clearness index. The available data, composed of normal incidence beam and global radiation measurements, have been analyzed on both a seasonal and a yearly basis. They have been compared to empirical equations previously reported in the literature for six sites in the Middle East region. In the case of Qidron, Israel, and three Greek sites, there was close agreement between the correlations, whereas in the remaining two sites (Gilat, Israel and Fudhaliyah, Iraq) the correlations were quite different. Reasons for the discrepancies in the latter two cases are presented in the text. The beam fraction of the global radiation was also correlated to the ratio of measured daily global to the corresponding clear sky daily global values, which were determined empirically in an attempt to reduce the degree of scatter observed in the correlations between beam fraction of global radiation and clearness index. This approach is intended to correct for the multiplicity of possible cloud conditions that can give the same value for the cloudiness index. It was not successful when applied to the Beer Sheva data, namely, it did not reduce the degree of scatter observed in the original correlations. This, in all likelihood, highlights the fact that the scatter is caused by local climatic conditions (such as turbidity and moisture content of the atmosphere).