Analysing alternative methods for estimating reference evapotranspiration (ET0) is essential for improving the management of water resources, especially where meteorological data are not fully available for application of the standard Penman-Monteith (PM) method. In this study, the aim was (a) to propose a model for estimating ET0 and (b) to investigate its application together with alternative methods in an area of fruit production in the semi-arid region of Brazil. The Souza-Silva model was generated to estimate ET0 and together with other original methods based on thermal-water variables and global incident solar radiation was tested relative to the PM method. The Souza-Silva method (ET0(ss)= - 0.00728 +1.356325DPV+ 0.174658Qg, where VPD is the water vapour-pressure deficit and Qg the global incident solar radiation) showed excellent performance in estimating ET0 with a low level of error (RMSE= 0.467 mm day(-1) and MBE= -0.034 mm day-1- ). Among the original alternative methods, those for Hargreaves (ET0(H)(s)) and Camargo (ET0(C)) models did not perform well in the region under study. The Benevides and Lopes (ET0(BL)) and Valiantzasl (ET0(V1)) methods gave excellent results; these were however even lower than methods based on the solar radiation. Compared to each of the original methods, the result of the Souza-Silva method showed greater precision and accuracy, with smaller estimation errors. It is concluded that this method, based on the water vapour-pressure deficit and solar radiation, can be used to estimate ET0 in places where data collection is limited.