Abstract Sedimentological data, in particular sediment grain size, from core MD95-2040 (2465 m water depth) and a nearby core (OMII-9K; 1833 m water depth), presently under the influence of Lower North Atlantic Deep Water and Deep Intermediate Water, respectively, have been used to reconstruct changes in the vigour of near-bottom water flow on the Iberian margin. The pattern of the palaeocurrent index indicates faster flow during the warm periods and slower in cold. This also holds in the Younger Dryas but there is an added effect of brief collapse of the circulation in Termination 1B due to meltwater input. The cold periods are generally characterised by elevated clay percentage (terrigenous material <2 μm diameter) and overall sediment accumulation fluxes. In comparison, warm periods of higher flow have lower depositional flux and a tendency to silt (terrigenous material 2–63 μm) dominance consistent with the higher flow speed and suppression of clay deposition. This is in contrast with a recent 230Th xs°-based study of two cores (MD95-2040 and -2039) [J. Thomson et al., Earth Planet. Sci. Lett. 165 (1999) 255–270], which concluded that increased glacial sediment and Th focussing was caused by stronger currents. The crucial factor is that the controlling speed is that of the flow at the depositional site rather than that of the erosional or transporting flow to the site. High glacial maximum fluxes of carbonate (∼6 g cm −2 ka −1) are largely composed of fine (<63 μm) non-coccolith material, suggesting a significant detrital component. It is proposed that the sediment and Th arrive on the lower slope via different routes, sediment down-slope in bottom nepheloid layers and turbidity currents, Th scavenged from the water column by organic matter, and are rapidly mixed on the bed. Removal of this material from a large area and its deposition in a small one adjacent to Oporto Seamount, together with a small amount of direct down-slope supply, accounts for the properties of the deposits. Allowance for erroneous fluxes due to possible coring expansion ‘stretching’ of the upper part of MD95-2040 does not invalidate these conclusions.