Transient evaporation method is promising for rapid measurement of soil water retention curve (WRC). Existing apparatuses of transient evaporation method are not ideal for measuring the WRC of structured soils because of the sample disturbance induced by invasive water content probes used. In this study, non-invasive TDR probes with different waveguide layouts (i.e., different number of conductors and waveguide length) were developed to measure the WRC of loess during transient evaporation, while the soil suction is recorded through a tensiometer. Results show that the WRCs of undisturbed and remolded loess measured by a three-conductor non-invasive TDR probe are in good agreement with the results of pressure plate tests. The differences in the air entry value (AEV) and desorption rate between the WRCs measured by the transient evaporation method and the pressure plate measurement are less than 10 %. On the other hand, the WRCs measured by a two-conductor non-invasive TDR probe underestimate the water content for a given suction. The AEV of undisturbed and remolded loess determined by the transient evaporation method is 66.7 % and 112.9 % smaller than the results of pressure plate measurements, respectively. The underestimation of the two-conductor non-invasive TDR probe in measuring the WRC of loess can be attributed to the fact that the vertical sampling range of two-conductor non-invasive TDR probe is two times greater than that of the tensiometer. The results in this study show the effectiveness of using the non-invasive TDR probe along with a tensiometer for measuring the WRC. More importantly, great caution should be taken on compliant sampling volume of the non-invasive TDR probe and tensiometer to obtain accurate WRC measurements.