Three megascopic and disseminated tephra layers (which we refer to as layers A, B, and C) occur in late Quaternary glaciomarine sediments deposited on the West Antarctic continental margin. The stratigraphical positions of the distal tephra layers in 28 of the 32 studied sediment cores suggest their deposition during latest Marine Isotopic Stage (MIS) 6 and MIS 5. One prominent tephra layer (layer B), which was deposited subsequent to the penultimate deglaciation (Termination II), is present in almost all of the cores. Geochemical analyses carried out on the glass shards of the layers reveal a uniform trachytic composition and indicate Marie Byrd Land (MBL), West Antarctica, as the common volcanic source. The geochemical composition of the marine tephra is compared to that of ash layers of similar age described from Mount Moulton and Mount Takahe in MBL and from ice cores drilled at Dome Fuji, Vostok and EPICA Dome C in East Antarctica. The three tephra layers in the marine sediments are chemically indistinguishable. Also five englacial ash layers from Mt. Moulton, which originated from highly explosive Plinian eruptions of the Mt. Berlin volcano in MBL between 142 and 92 ka ago, are chemically very similar, as are two tephra layers erupted from Mt. Takahe at ca 102 ka and ca 93 ka. Statistical analysis of the chemical composition of the glass shards indicates that the youngest tephra (layer A) in the marine cores matches the ash layer that erupted from Mt. Berlin at 92 ka, which was previously correlated with tephra layers in the EPICA Dome C and the Dome Fuji ice cores. A tephra erupted from Mt. Berlin at 136 ka seems to correspond to a tephra layer deposited at 1733 m in the EPICA Dome C ice core. Additionally, the oldest tephra (layer C) in the marine sediments resembles an ash layer deposited at Vostok around 142 ka, but statistical evidence for the validity of this correlation is inconclusive. Although our results underscore the potential of tephrostratigraphy for correlating terrestrial and marine palaeoclimate archives, our study also reveals limitations of this technique, which may result in the miscorrelation of tephra. Such pitfalls comprise failure to recognise the occurrence of various tephra layers in marine sediment cores, `swamping' of records with chemically indistinguishable tephra from a single volcanic source, and exclusive use of `geochemical fingerprinting' for correlating ash layers. (C) 2007 Elsevier Ltd. All rights reserved.