We implemented the Amenu-Kumar model [Amenu and Kumar, 2008] to simulate plant root hydraulic redistribution and analyzed its impacts on global evapotranspiration (ET) in the Community Land Model (CLM4.5). We tested two numerical implementations: the first solved the coupled root water transport equation and the soil water Richards' equation simultaneously, while the second solved the two equations sequentially. We demonstrate that the sequential implementation is numerically incorrect, whereas the tightly coupled implementation is numerically robust with numerical time steps varying from 1 to 30 minutes. However, we found the sequential implementation resulted in better comparison to measurements at the Blodgett Forest site in the Sierra Nevada range, California. For annual global land ET, the sequential implementation overestimates by as much as 5 mm day-1 in some grid cells with root hydraulic redistribution rates compared to that from the tightly coupled implementation. Globally, we found including root hydraulic redistribution resulted in the CLM4.5 predicted ET agreeing better with the FLUXNET-MTE ET predictions north of 20å¡ N and agreeing worse in tropical regions, no matter which pedotransfer function or meteorological forcing we applied. In addition, we found including root hydraulic redistribution did not completely resolve (but did alleviate) the ET enhancement in vegetation removal experiments. Based on our analysis, we recommend modifications for future versions of CLM to improve the simulation of ET.