Data of a field experiment carried out at Petrolina-PE, a semi-arid region of northeast Brazil, were used to evaluate the evapotranspiration during the 1999 fruiting cycle of a mango orchard. Sensors for net radiation, two levels dry and wet bulbs air temperature and wind speed were installed above a mango tree in a micrometeorological tower. Soil heat flux sensors were installed at 2 cm below soil surface and soil moisture content was measured by sets of tensiometers installed each at 20 cm from the surface down to 220 cm soil depth. The individual mango tree evapotranspiration was obtained by two methods: Bowen ratio-energy balance (BREB) and soil water balance (SWB). Daily mango orchard evapotranspiration increased slowly from 3.1 mm per day at the beginning of the experimental period (middle July) to 4.9 mm per day at the maximum growth period of the fruit. Then, it decreased to reach a 4.1 mm per day value, approximately at the full maturation fruit. The accumulated mango orchard water consumption for the whole productive cycle was 551.6 and 555.1 mm by the soil water and Bowen ratio-energy balance methods, respectively. For the experimental climate conditions, the soil water balance method was shown to be more efficient in the measurement of the mango orchard daily evapotranspiration when compared to the energy balance method, particularly when obtained for time period equal to or greater than 7 days, with a percentage error of estimation of daily crop evapotranspiration around 7%. Also, in the study region, the mango orchard water requirements are not constant throughout the productive cycle. However, it can be estimated by the crop coefficient ( K c) obtained as a function of the days after flowering (DAF) as: K c=0.36+0.009(DAF)−4×10 −5(DAF) 2. On average, the percentage error associated to the estimation of the latent heat flux increased from 17% for low to 38% for high atmospheric energy demand with intermediate values, around 25%, for moderate available atmospheric energy.