This paper reports on the use of a multi-regression model adapted at the University of Reunion Island, that was formulated for surface temperature trend estimates in South Africa. Depending on the location of the observing site, the selected geophysical signal is expressed in the form of a sum of forcings, which account for most of its variability. The trend values are then derived from the residual terms as a linear function. The atmospheric forcings included in the model are: Annual, Semi-annual, Quasi-Biennial Oscillation, El-Nino Southern Oscillation (ENSO), the 11-year solar cycle (SSN) and Indian Ocean Dipole (IOD). Long-term databases of more than 20 years are typically considered for the measurement of decadal trends and to determine the contribution of different atmospheric forcings. In this study, the surface minimum and maximum temperature datasets collected from 1980 to 2004 (25 years) at the South African Weather Service (SAWS) Addo station (33.568°S; 25.692°E), located in the Eastern Cape, South Africa are employed. The first results obtained for this station are presented and illustrate the validity of the model to account for observed trends.