Abstract: Bankfull discharge is morphologically significant, as it is identified as a key parameter for studying river morphology, sediment motion, flood dynamics, and their ecological impacts. Identification and determination of bankfull stage is critical to both development of regional curves and calculation of bankfull discharge but doing this in-situ could be practically difficult and subjective. This paper compares two identifying approaches, one based on geomorphic features in-situ, and the other on geometric criterion of surveyed channel cross-sections. For the geometric criterion, we collected the data of cross sections measured at the Huayuankou hydrology station in 50 years to compare its four methods, namely Wolman (WOL), Williams (WIL), Riley (RIL) and Wang (B/H). Analysis shows that to identify the bankfull stage in a multiple-branch wandering reach, effective channel width is a necessary parameter even all the cross-sectional areas below various stages are known. Of these four methods, WIL and RIL are insensitive to effective width, while the bankfull stages estimated by WOL and B/H tend to increase with narrower effective width. In application to the cross sections of Huayuankou, the bankfull stages by WIL are relatively higher, especially those for the period after 1980, while the stages by WOL and B/H are relatively lower.