<正>Some concepts related to sediment transport in rivers with overbank flow are described.Following a description of the physical processes that are involved when a river inundates its floodplains,some simple com- putational methods are presented which permit the depth-averaged velocity and boundary shear stress to be pre- dicted within a cross section of variable,but prismatic shape.The methoda account for the strong transverse shear in velocity that occurs when the stage is just above bankfull,as well as secondary flows and vertical shear. Data from laboratory experiments are used to illustrate how overbank flows affect the sediment transport rate.It is shown,for example,that floodplain roughness can have a significant effect on conveyance capacity, division of flow between a river channel and its associated floodplains,resistance,bed-load transport and dune migration rate.Although much may be gained from such small scale experiments,typically conducted under steady flow conditions,flood flows in natural rivers are known to be much more complex.This is due to the un- steadiness in the flow,the presence of vegetation in the channel or on the floodplain,and the significant varia- tion that there usually is in channel and floodplain topography.Despite this,the modelling of sediment transport in rivers is still mainly based on empirical equations derived from steady inbank flow data.The overbank flow data presented herein may challenge this approach and help to improve modelling techniques for certain types of river. One particular approach to modelling sediment transport in rivers with overbank flow is based on the so- called“Shiono and Knight method”(SKM).The basis of this method will be briefly explained and some appli- cations presented.The SKM forms part of the“Conveyance Estimation System”(CES) software,recently de- veloped by HRWallingford and adopted by the Environment Agency for use in the UK.The CES (www.river- conveyance,net) is a tool for determining the stage-discharge relationship for flood flows in rivers via a con- veyance generator,a roughness advisor and an uncertainty estimator.