Six years of hourly global and diffuse irradiation measurements on a horizontal surface performed at Athalassa, Cyprus, are used to establish a relationship between the daily diffuse fraction and the daily clearness index. Two types of correlations-yearly and seasonal-have been developed. These correlations, of first and third order in the clearness index are compared to the various correlations established by Collares-Pereira and Rabl (1979), Newland (1989), Erbs et al. (1982), Rao et al. (1984), Page (1961), Liu and Jordan (1960) and Lalas er al. (1987). The comparison has been performed in terms of the widely used statistical indicators (MBE) and (RMSE) errors; an additional statistical indicator, the t-statistic, combining the earlier indicators, is introduced. The results indicate that the proposed yearly correlation matches the earlier correlations quite closely and all correlations examined yield results that are statistically significant. For large (K) over bar(T)>0.60 values, most of the earlier correlations exhibit a slight tendency to systematically overstimate the diffuse fraction. This marginal disagreement between the earlier correlations and the proposed model is probably significantly affected by the clear sky conditions that prevail over Cyprus for most of the time as well as atmospheric humidity content. It is clear that the standard correlations examined in this analysis appear to be location-independent models for diffuse irradiation predictions, at least for the Cyprus case.