The 1979 Nice turbidity current is modelled using a visco-plastic analysis of flow velocity because the initial flow concentrations are expected to have been very high. The complete history of the failed sediment from debris flow to turbidity current plume is therefore addressed. The turbidity current portion is considered as a steady state flow divided into a dense bottom flow and an upper plume. Model results show that a dense flow can be generated from the debris flow by the disaggregation of the initial slide. The dense flow would be strongly erosive and able to create and maintain a low-density plume at its surface. The depth of erosion of the channel floor by the dense flow is predicted to reach 6–11 m in overconsolidated sediments, with the main erosion taking place in Var Canyon and the Upper Fan Valley. The eroded volume (108 m3) provides additional material to the sediment mass of the initial failure. The dense flow appears able to inject fine sand and silt into the overlying plume during 90 km, and would disintegrate before being able to deposit sediment. The extensive sand layer along the travel path of the turbidity current may have been deposited from the tail of the trailing plume: a result of the velocity difference between the plume and the dense flow. Observations on sedimentary structures, erosion features and distribution of the sand deposit are quite in agreement with our modelling approach. For example, gravel waves can be generated when loose deposits are reworked by the supercritical dense flow. The methodology and equations presented here provide a good estimate of the geological consequences of a high-velocity gravity flow undergoing rheological transition.