The fractionation of 230Th and 231Pa was investigated throughout the Atlantic sector of the Southern Ocean. Published scavenging models generally assume that the 231Pa/230T ratio of surface sediments is primarily determined by the mass flux of particles. This relationship holds north of the Polar Front, where low primary productivity coincides with ratios of unsupported 231Pa/230Th-xs(231Pa/230Th) - in surface sediments below the production ratio of both radionuclides in the water column. However, we observed high xs231Pa/230Th ratios, conventionally interpreted as a high-productivity signal, in surface sediments south of the Polar Front, especially throughout the Weddell Sea, in contradiction with the low particle flux of this region. Measurements of both dissolved and particulate fractions of 231Pa and 230Th in the water column revealed a strong N-S decrease in the Th/Pa fractionation factor, from typical open ocean values around 10 north of the Polar Front to values between 1 and 2 south of 60°S. This observation clearly indicates that the high xs231Pa/230Th ratios in surface sediments south of the Antarctic Circumpolar Current are produced by a N-S increase in the relative scavenging efficiency of 231Pa relative to 230Th, most probably due to a change in the chemical composition of particulate matter, and not by a high mass flux. It is speculated that biogenic opal, suggested not to significantly fractionate231Pa and 230Th, may explain the enhanced scavenging of 231Pa to the south. This assumption is further supported by extremely high 231Pa/230Th ratios up to 0.34 in material collected with sediment traps south of the Polar Front, where fluxes are primarily determined by biogenic opal. Based on these results we conclude that, in regions where the sedimenting flux is dominated by biogenic opal, the 231Pa/230Th ratio is not a reliable indicator for the mass flux of particles, thus limiting its use as a paleoproductivity proxy in the Southern Ocean.