The probabilistic speed-frequency distribution of the wind is essential for evaluation of the wind potential. Although wind commonly evaluated with weibul distribution, by assigning null value to calm winds it cannot envisage the existence of calm winds. Since for the sites with significant calm wind frequency Weibull distribution is uncertain, in this paper we present a new theoretical approach which employs Maximum Entropy Principle (MEP). The model is improvement of previous proposed MEP which applied to the synoptic sites distributed inside the Tunisian territory with significant calm winds frequency. The obtained results appropriately describe the distribution of measured wind speed data particularly calm speed over the MEP and Weibull models.