Many of the world's flood basalt provinces form elevated plateaux at the margins of continents, although in most cases their present large elevation is not the result of mountain building processes. Several explanations have recently been put forward to explain such occurrences of epeirogeny. The Deccan Trap basalt province forms one such elevated plateau, and results are presented here showing how the epeirogenic uplift in this region, combined with crustal subsidence probably associated with the rifting of the Indian continental margin, has affected the structure of the basalt sequence. Trace element analytical data are used for samples from numerous vertical sections through the Deccan Traps lava series along and around the Western Ghats ridge in India. The results reinforce the previously defined stratigraphy of the Mahabaleshwar area, and extend it over a region covering some 36 000 km2, reaching as far south as Belgaum and the Trap/basement contact. These results show that the lava pile is not flat lying, but forms a very low amplitude anticlinal fold structure plunging southwards by up to 0.3 ° over most of the area, although in the south there is evidence of a reversal of this plunge. The fold is interpreted as being the result of two tilting processes: (1) westward tilting near the coast, due to the foundering of the passive continental margin, and (2) epeirogenic uplift along the whole west coast of India producing the observed topography and the peninsula-wide drainage patterns, and also the easterly component of dip. Variations in the magnitude of the latter effect along the western continental margin may also be important in generating the plunge of the fold, although the possibility of some component of depositional dip may also be important. This latter possibility can be modelled using a simple computer program. The results of this modelling show that a migrating linear volcanic edifice fits the observations best.