The relationship between terrestrial irradiance and sunshine duration is not crisp and there are scatters around a general trend, which most often is expressed to occur in the form of a linear expression. This study presents a way of grouping the solar irradiation-sunshine duration data into convenient seasonal subgroups and then makes predictions within each of the groups quantitatively. In the classical Angstrom or other approaches, the seasonal variations are not considered, and therefore, rather global parameter estimations are obtained. However, the seasonal methodology of this paper provides more detailed interpretations in addition to seasonal effects and parameter estimations. (C) 2002 Elsevier Science Ltd. All rights reserved.