Abstract Shelf-margin or shelf-edge deltas are a common constituent of Quaternary shelves, where their genetic link to falling and lowstand sea level has been well established, but they are rarely reported from older successions. Recognition of this delta type in the ancient record is nevertheless important, because (1) it helps to position the fossil shelf edge, (2) it provides insight into how sand budgets were partitioned between shelf edge, slope and basin-floor settings, and (3) it can contribute to the debate concerning three versus four systems-tract systematics and the positioning of a sequence boundary. Shelf-margin deltas create wide (tens of kilometers), high (hundreds of meters) and steep (3–6°) clinoforms, because they build across the relatively deepwater shelf margin. Such bodies tend to form strike-elongate wedges that initially thicken basinwards, attaining a maximum thickness of 50–200 m just outboard of the shelf edge, then thin on the middle/lower slope, whereas they commonly pinch out landwards by lapping back onto shelf shales. Prior to reaching the shelf edge, small-scale (tens of meters), tangential foresets (usually The key facies association in shelf-edge deltas is the mouth bar-to-delta front association. Mouth-bar facies, landwards of the shelf edge, consist mainly of thick, clean, flat to low-angle and ripple-laminated medium to fine sands. Basinwards from the shelf edge, such sands commonly alternate with heterolithic slumped unit, creating characteristic slumped to laminated couplets. Because the delta front is superimposed on a preexisting, steep and extended shelf margin, it commonly contains (beyond the mouth bars) thick successions (up to many tens of meters) of sandy, slope turbidites. Shelf-margin deltas differ from inner shelf deltas in showing: (1) an order of magnitude higher clinoforms (hundreds instead of tens of meters), and strike-elongated, locally pod-like sand bodies commonly affected and augmented by growth faulting; (2) paleoecological evidence of abrupt shallowing (foreshortened stratigraphy); (3) turbidite-prone delta-fronts; (4) larger scale and greater abundance of slope-controlled soft-sediment deformation and (5) the general absence of a paralic ‘tail’ along the trailing edge of the delta front. Shelf-margin deltas can be classified into two types. Stable shelf-margin deltas are usually tens of meters thick and are not associated with shelf-edge incision or major slope collapse/disruption features. Slope turbidites are common and occur as unconfined sheets and lobes. Such deltas form when relative sea level falls no lower than the level of the shelf platform or, after longer sea-level falls, when rivers reestablish at the shelf edge on the rise. Unstable shelf-margin deltas are associated with large-scale slope collapse, listric growth faults and sometimes with salt diapirs. Slides are common on the upper slope, whereas imbricated sediment packages and compressional ridges affect the slope toe. Slope turbidites can be ponded within structurally controlled mini-basins on the slope. Such deltas are often related to unusually great sand influx to the shelf edge, or simply to prolonged or large fall of relative sea level below the shelf edge. The diachronous erosional unconformity atop the shelf-to-shelf-edge deltaic wedge is viewed by some researchers as the most easily recognizable and persistent surface on the shelf and slope. This surface contrasts with the downlap erosional surface developed from the beginning of sea-level fall, which is viewed by other researchers as the key boundary surface within this complex. It should be noted that sea level can fall for a long period of time (tens of thousands of years) before deltas even reach the shelf margin and, therefore, before significant volumes of sand are delivered across the shelf break. Hence, the time of incision of the shelf edge and emplacement of deepwater sand is commonly long after the initial fall so the time of (maximum) relative lowstand of sea level may be a more practical choice for the timing of the sequence boundary. Recognition of shelf-margin deltas and analysis of their architecture help in the prediction of presence or absence of basin-floor fans. Deltaic complexes that are aggradational to backstepping, downlap onto disrupted or complex slopes, and overlie an incised shelf and shelf edge, predict that there should be basin-floor fans present. In contrast, delta complexes that show a prolonged and preserved progradational to downstepping architecture implies the presence of turbidite accumulations on the slope but not on the basin floor.