The effectiveness of wind water pumping plant depends on the available wind potential in the region and on the plant components sizing. This paper presents an algorithm for wind potential assessment based on the widely used Weibull distribution. As many methods are adopted to determine Weibull parameters, an improvement version based on the selection of the most accurate method and the establishment of a huge database using an artificial neural network (ANN) is proposed. Since the site wind performance is evaluated, the wind generator blades surface is computed on the basis of the variation limits of the monthly wind potential and the well height of rise. The sizing principal considers the calculation of the gravity centre of the general function of surface. Results are illustrated using meteorological database provided by the National Institute of Meteorology (INM) corresponding to Sfax, Tunisia. Obtained results confirm that the modified maximum likelihood method (MMLM) is the most accurate one as it provides a monthly error between À11:6% and 2:3%. Hence, a typical pumping plant, with monthly water need of 15 m 3 month located in Sfax, Tunisia, requires 37 m 2 as optimum blades surface.