In this study, several Angstrom-type regression models, namely the linear, quadratic, cubic, logarithmic and exponential models, are compared to estimate the global solar radiation on horizontal surfaces over a territory. The country of Tunisia is used as a case study for the methodology. The models studied were calibrated using four meteorological stations that are providing global solar radiation and sunshine duration data. The statistical analysis showed that the models assessed are well suited to accurately estimate the solar potential. Overall, the cubic model showed the best regression fit and performed slightly better than the others. However, since the difference between the models studied is insignificant, the linear model has been preferred for the mapping of the solar potential of the study area, mostly because of its simplicity. Two approaches were assessed for the use of the Angstrom-Prescott parameters: the classical approach which consists of using generic parameters for the whole study area, and a spline interpolation of these parameters. Both methods gave very similar results with a slight improvement for the classical approach which has been employed to map the spatial and temporal distribution of the solar potential of the study area using spline interpolations at a 500 m x 500 m resolution. In application to the territory of the case study, the monthly mean daily global solar radiation in Tunisia varies between 6.9 MJ/m(2)/day in the north of the country during January and 28 MJ/m(2)/day in June, in the surroundings of the Gulf of Gabes. Annually, the monthly mean daily solar radiation ranges from 15.7 to 19.4 MJ/m(2)/day, increasing from north to south. (C) 2015 Elsevier Ltd. All rights reserved.