Understanding the variation and magnitude of crop coefficient (Kc) is important for accurate determination of crop evapotranspiration and water use. In this study, we calculated Kc in an irrigated maize field with ground mulching by eddy covariance evapotranspiration measurements during the whole growing periods in 2009 and 2010 in an arid region of northwest China. A semi-empirical practical approach for estimating Kc was proposed by introducing the dynamic fraction of canopy cover and incorporating the effect of leaf senescence as a function of days after sowing. The contribution of arid advection of sensible heat resulting from irrigation to Kc and the response of Kc to canopy conductance (Gc) were investigated. The averaged values of daily Kc were lower than typical values obtained previously without mulching due to decreasing effect of mulching on Kc, with 0.82 and 0.80 for the 2 years, respectively. The maximum average Kc occurred at the heading stage, with 1.21 and 1.04 for the 2 years, respectively. The difference of Kc was attributed to the difference of leaf area index. The semi-empirical practical approach could well estimate the variations of Kc, thus could be a robust and useful tool for the practical users and water managers. The contributions to daily Kc from the arid advection were 4.4–28.0% of the measured Kc. The Gc had stronger control on daily Kc at the early and later stages than at the middle stage. When Gc, leaf area index and relative soil extractable water were lower than the respective threshold values of 20 mm s−1, 3.0 m2 m−2 and 0.5, the daily Kc increased significantly with the increase of the three factors, and almost remained constant when the three factors were beyond the threshold values. These results are helpful for quantifying contributions of individual factors to Kc, and subsequently improving water management practices according to Kc. Copyright © 2014 John Wiley & Sons, Ltd.