Drip irrigation (DI) has been widely utilized for crops and its water-saving effect has been confirmed by numerous studies. However, whether this technology can save so much water under the field scale during practical application is still uncertain. In order to answer this question, evapotranspiration (ET), soil water content, transpiration and evaporation over the DI and border irrigation (BI) in an arid area of NW China were continuously measured by two eddy covariance systems, micro-lysimeters, the packaged stem sap flow gauges and CS616 sensors during 2014–2018 growing seasons. The results showed that the DI averagely increased crop water use efficiency (CWUE) by 11% per year against BI. The deep drainage under DI treatment was lower than BI by 8% averagely for the five-year period. While for the ET, the DI averagely decreased ET by 7% and 40mm per year against the traditional BI. The decrease in ET was mainly due to the significant reduction in soil evaporation instead of transpiration. Oppositely, we found that DI may increase maize (Zea mays L.) transpiration in some year for the better preponderant growth of crop. Thus, the accelerating effect on transpiration of DI and its reducing effect on soil evaporation should be considered simultaneously. In our experiment, DI only improved CWUE and WUE (water use efficiency) by 11% and 15% on average in a large farmland scale, unable to always be more than a 20% improvement, as concluded by many other field experiments. Consequently, the water-saving effect of DI should not be overestimated in water resource evaluation.