Submarine-fan channels can build the largest sediment accumulations on Earth, but our understanding of flow and sedimentation processes related to channel evolution remains limited. Results from physical and numerical modelling predict dominantly downstream channel bend migration. However, observations and evolutionary models for aggradational submarine channels on passive margins suggest that bends are dominated by lateral expansion. This paradox may be due to limitations induced by the use of constant width channels in process studies. Constant width has been used for two reasons: partly because this is the simplest possible case, but primarily because the width variation around submarine channel bends is unknown. Channel width variations are examined from an active channel reach with 49 bends and three inactive but unfilled channel reaches with a total of 35 bends from the Congo Fan. Each bend was divided into 13 cross-sections, and for each cross-section, channel width was measured for the channel base, and at 10 m vertical increments up to the height of the channel banks. The results indicate that channels are typically wider around bend apices than around inflections. We argue that this morphology suggests that channels are controlled by bank-pull (outer bank erosion), with later deposition at the inner bend, similar to many rivers. The implications of these spatial changes in channel width around bends for sedimentation and channel evolution are explored, and we suggest that such changes may account for the contradictions between physical and numerical modelling, and seafloor observations. Integration of these channel width data with the known climate history of the Congo Fan, further suggests that the magnitude of channel width variation at bend apices may be controlled by allogenic forcing, with larger flows associated with greater width variations around bends.