Streambed horizontal hydraulic conductivity (Kh) has a substantial role in controlling exchanges between stream water and groundwater. We propose a new approach for determining Kh of the shallow streambed sediments. Undisturbed sediment samples were collected using tubes that were horizontally driven into streambeds. The sediment columns were analysed using a permeameter test (PT) on site. This new test approach minimizes uncertainties due to vertical flow in the vicinity of test tube and stream stage fluctuations in the computation of the Kh values. Ninety-eight PTs using the new approach were conducted at eight sites in four tributaries of the Platte River, east-central Nebraska, USA. The Kh values were compared with the nondirectional hydraulic conductivity values (Kg) determined from 12 empirical grain-size analysis methods. The grain-size analysis methods used the same sediment samples as Kh tests. Only two methods, the Terzaghi and Shepherd methods, yielded Kg values close to the Kh values. Although the Sauerbrei method produced a value relatively closer to Kh than other nine grain-size analysis methods, the values from this method were not as reliable as the methods of Terzaghi and Shepherd due to the inconsistent fluctuation of the average estimates at each of the test sites. The Zunker, Zamarin, Hazen, Beyer, and Kozeny methods overestimated Kh, while the Slichter, US Bureau of Reclamation (USBR), Harleman, and Alyamani and Sen methods underestimated Kh. Any of these specific grain-size methods might yield good estimates of streambed Kh at some sites, but give poor estimates at other sites, indicating that the relationship between Kg and Kh is significantly site dependent in our study. Copyright © 2011 John Wiley & Sons, Ltd.