Abstract: This study tested the ability of four published pedotransfer functions (PTFs) (including ROSETTA, RAWLS, CAMPBELL and VAUCLIN) to predict unsaturated hydraulic conductivity for 56 sandy soil samples from the UNSODA database. Different PTFs for the estimation of saturated hydraulic conductivity were also evaluated in this study. Results showed that among the four PTFs, VAUCLIN produced the best performance, followed by ROSETTA and RAWLS, and CAMPBELL gave the worst performance when saturated hydraulic conductivity predicted by the identical PTFs as input. The performance of RAWLS, CAMPBELL and ROSETTA has been improved to different extent when measured saturated hydraulic conductivity as input. However, VAUCLIN did not produce the best performance when measured saturated hydraulic conductivity as input. The influence of saturated hydraulic conductivity on the predictive accuracy of unsaturated hydraulic conductivity was significant. Overall, CAMPBELL gave the best performance when the measured saturated hydraulic conductivity was used in the functions for estimation of unsaturated hydraulic conductivity.