This paper reports the results of using three empirical methods (Makkink, Priestley-Taylor and Hargreaves) for estimating the reference evapotranspiration (ET0) in the semi-arid region of Tensift Al Haouz, Marrakech (center of Morocco). The Penman-Monteith equation, standardized by the Food and Agriculture Organization (FAO-PM), is used to evaluate the three empirical methods. The obtained ET0 data were used to estimate crop water requirement (ET) of winter wheat using the crop coefficient (K-c) approach and results were compared with ET measured by the Eddy Covariance technique. The result showed that using the original empirical coefficients a, alpha and C-m in Hargreaves, Priestley-Taylor and Makkink equations, respectively, the Hargreaves method agreed fairly well with FAO-PM method at the test site. Conversely, the Priestley-Taylor and Makkink methods underestimate the ET by about 20 and 18 %. After adjustment of the original values of two parameters alpha and C-m coefficients in Priestley-Taylor and Makkink equations, the underestimation of ET was reduced to 9% and 4% for the Priestley Taylor and Makkink methods, respectively, which led to an improvement of 55% and 76% of the obtained values compared with the original values.