In 2004, an aerial reconnaissance survey of Gulf of Carpentaria rivers identified alluvial gully erosion as a likely key sediment source into many rivers. The process is found to varying degrees within alluvial river types along most Gulf rivers, and is more extensive in the larger, steeper gradient rivers. It is hypothesised that the high connectivity between alluvial gullies and trunk streams makes these features a significant sediment source to the Gulf. The process appears to differ significantly in scale and process from the well documented, largely colluvial, gullies that abound in southern Australia. New conceptual models of the processes driving these gullies, and controls on their morphology and spatial distribution, are required if this process is to be adequately parameterised into existing sediment budget models, such as SedNet, in northern Australia. It is likely that similar features exist in eastern draining tropical rivers, suggesting existing SedNet models may need to be adjusted to account for this different style of gully erosion. Our model suggests that morphological variability can largely be explained by the relative dominance of the two main processes driving gully headward retreat: basal sapping and overland flow driven knickpoint retreat. The model suggests that gully initiation, rate of activity and morphological variability can be explained by the interplay between soil type, floodplain relief, vegetation, climate, fire regime, grazing pressure, river flow regime and local rainfall within the context of these two primary driving processes.