This study proposes that the submarine canyons of the northern Kwazulu-Natal continental margin formed contemporaneously with hinterland uplift, rapid sediment supply and shelf margin progradation during the forced regression of upper Miocene times. These forced regressive systems tract deposits volumetrically dominate the shelf sediments, and comprise part of an incompletely preserved sequence, amongst which six other partially preserved sequences occur. The oldest unit of the shelf corresponds to forced regression systems tract deposits of Late Cretaceous age (seismic unit A), into which a prominent erosive surface, recognized as a sequence boundary, has incised. Fossil submarine canyons are formed within this surface, and underlie at least one large shelf-indenting canyon in the upper continental slope. Smaller shelf indenting canyons exhibit similar morphological arrangements. Late Pliocene deposits are separated from Late Cretaceous lowstand deposits (seismic unit B) by thin veneers of Late Palaeocene (seismic unit C) and mid to early Miocene (seismic unit D) transgressive systems tract deposits. These are often removed by erosive hiatuses of early Oligocene and early to mid Pliocene age. These typically form a combined hiatus surface, except in isolated pockets ofthe upper slope where late Miocene forced regressive systems tract units are preserved (anomalous progradational seismic unit). These sediments correspond to the regional outbuilding of the bordering Tukhela and Limpopo cones during relative sea level fall. Either dominant late Pliocene sediments (seismic unit E), or transgressive systems tract sediments which formed prior to the mid Pliocene hiatus, overlie these sediments. Widespread growth faulting, slump structures and prograding clinoforms towards canyon axes indicate that these sediments initiated upper slope failure which served to create proto-canyon rills from which these canyons could evolve. The association of buried fossil canyons with modern day canyons suggests that the rilling and canyon inception process were influenced by palaeotopographic inheritance, where partially infilled fossil canyons captured downslope eroding flow from an unstable upper slope. Where no underlying canyons occur, modern canyons evolved from a downslope to upslope eroding system as they widened and steepened relative to the surrounding slope. Statistical quantification of canyon forms shows a dominance of upslope erosion. Landslide geomorphology and morphometric analysis indicate that this occurred after downslope erosion, where the canyon axis was catastrophically cleared and incised, leading to headward retreat and lateral excavation of the canyon form. Trigger mechanisms for canyon growth and inception point to an overburdening ofthe upper slope causing failure, though processes such as freshwater sapping may emulate this pattern of erosion. It appears that in one instance, Leven Canyon, freshwater exchange with the neighbouring coastal waterbodies has caused canyon growth. The canyons evolved rapidly to their present day forms, and have been subject to increasingly sediment starved conditions, thus limiting their evolution to true shelf breaching canyon systems. Sedimentological and geomorphological studies show that the shelf has had minor fluvial influences, with only limited shelf-drainage interaction having occurred. This is shown by isolated incised valleys of both Late Cretaceous and Late PleistocenelHolocene age. These show classic transgressive valley fills of wave dominated estuaries, indicating that the wave climate was similar to that of today. The narrowness of the shelf and the inheritance of antecedent topography may have been a factor in increasing the preservation potential of these fills. Canyons thus appear to have been "headless" since their inception, apart from Leven Canyon, which had a connection to the Last Glacial Maximum (LGM) St Lucia estuary, and Wright Canyon, which had an ephemeral, shallow LGM channel linking it to the Lake Sibaya estuarine complex. Coastline morphology has been dominated by zeta bays since at least 84 000 BP, thus littoral drift has been limited in the study area since these times. The formation of beachrock and aeolianite sinks during regression from the last interstadial has further reduced sediment supply to the shelf. The prevalence of sea-level notching in canyon heads, associated with sea levels of the LGM indicates that canyon growth via slumping has been limited since that time. Where these are obscured by slumping in the canyon heads (Diepgat Canyon), these slumps have been caused by recent seismic activity. The quiescence of these canyons has resulted in the preservation of the steep upper continental slope as canyon erosion has been insufficient to plane the upper slope to a uniform linear gradient such as that of the heavily incised New Jersey continental margin. Progressive sediment starvation of the area during the Flandrian transgression has resulted in a small shore attached wedge of unconsolidated sediment (seismic unit H) being preserved. This is underlain by a mid-Holocene ravinement surface. This crops out on the outer shelf as a semi-indurated, bioclastic pavement. Thinly mantling this surface are Holocene sediments which have been reworked by the Agulhas Current into bedforms corresponding to the flow regime and sediment availability to the area. Bedforms are in a state of dis-equilibrium with the contemporary hydrodynamic conditions, and are presently being re-ordered. It appears that sediment is not being entrained into the canyons to the extent that active thalweg downcutting is occurring. Off-slope sediment loss occurs only in localized areas, supported by the dominance of finer grained Early Pleistocene sediments of the outer slope. A sand ridge from the mid shelf between Wright and White Sands Canyons appears to have been a palaeo-sediment source to White Sands Canyon, but is currently being reworked southwards towards Wright Canyon. The prevalence of bedform fields south of regularly spaced canyon heads is considered a function of hydrodynamic forcing of the Agulhas Current by canyon topography. These bedforms are orientated in a northerly direction into the canyon heads, a result ofnortherly return eddying at the heads of these canyons. / Thesis (Ph.D.)-University of KwaZulu-Natal, Westville, 2009.