This article reports the results of evapotranspiration (ET) experiments carried out in Southern Italy (Sicily) in a pilot-scale constructed wetland (CW) made of a combination of vegetated (Phragmites australis) and unvegetated sub-surface flow beds. Domestic wastewater from a conventional wastewater treatment plant was used to fill the beds. Microclimate data was gathered from an automatic weather station close to the experimental plant. From June to November 2009 and from April to November 2010, ET values were measured as the amount of water needed to restore the initial volume in the beds after a certain period. Cumulative reference evapotranspiration (ET(0)) was similar to the cumulative ET measured in the beds without vegetation (ET(con)), while the Phragmites ET (ET (phr) ) was significantly higher underlining the effect of the vegetation. The plant coefficient of P. australis (K(p)) was very high (up to 8.5 in August 2009) compared to the typical K(c) for agricultural crops suggesting that the wetland environment was subjected to strong "clothesline" and "oasis" effects. According to the FAO 56 approach, K(p) shows different patterns and values in relation to growth stages correlating significantly to stem density, plant height and total leaves. The mean Water Use Efficiency (WUE) value of P. australis was quite low, about 2.27 g L(-1), probably due to the unlimited water availability and the lack of the plant's physiological adaptations to water conservation. The results provide useful and valid information for estimating ET rates in small-scale constructed wetlands since ET is a relevant issue in arid and semiarid regions. In these areas CW feasibility for wastewater treatment and reuse should also be carefully evaluated for macrophytes in relation to their WUE values.