A process-product study of flow processes and sedimentation along the entire length of a sinuous deep-water channel in Yinggehai Basin yields three major differences in channel morphometrics, architecture, and evolution, each of which is related to a specific flow property estimated from channel morphology. Firstly, downstream channels contain inner-bend deposits, whereas upstream channels do not, which is ascribed to the downstream decrease in water entrainment (Ew) and peak velocity (up). The downstream decrease in Ew and up are shown to enhance stratification, which, in turn, resulted in the secondary flow reversal at the bends and ultimately gave rise to the occurrence of inner-bend deposits. Secondly, downstream channels have ca. 10–20 × more swing and 5–6 × lateral migration than upstream channels, which is ascribed to the downstream decrease in flow velocity (U) and Froude number (Fr). Turbidity flows in upstream channels were supercritical (Fr of 1.383 to 1.648) and bypassing dominated, and had high U of 1.7 to 2.6 m/s. They tended to inhibit deposition but to promote downcutting, and hence correspond to an architectural style of vertical aggradation and an evolution pattern of downdip sweep (down system meander loop migration). Turbidity currents in downstream channels, in marked contrast, were subcritical (Fr of 0.914 to 0.998) and weakly bypassing, and had low U of 1.2 to 1.3 m/s, all of which promoted deposition but disfavored channel thalweg erosion, thus encouraging an architectural manifestation of lateral migration and an evolution model of lateral swing (meander loop expansion). Thirdly, downstream channels present overbank deposits, whereas upstream ones do not, which could be explained by the bankfull discharge (Qc) decrease of turbidity currents downstream. The magnitude of Qc decrease downstream is great, far exceeding the descent degree of Ew down the channel, and, thus, we can get the deduction that channel-forming flows were over spilling downstream and led to the occurrence of levee deposits.