The Himalaya are the physical response to collision between the Indian subcontinent and Eurasia, which commenced in the Paleocene. They have been considered for many years to represent the type-example of collisional orogenesis. However, despite numerous detailed studies of the petrology, geochemistry and structure of the igneous and metamorphic components, there is no clear consensus or simple, integrated model to explain the thermal history of the deepest rocks in the orogen, the formerly partially molten Greater Himalayan Sequence (GHS). We believe that this is a result of the emphasis on the early Miocene stage of high-temperature metamorphism and extensive melting which has nearly obliterated an earlier stage of pre-Oligocene high-pressure metamorphism. The cause of melting is enigmatic, and is commonly ascribed to either shear heating, or unusually high concentrations of radioactive elements accompanying burial to depths around 35 km. However, we believe the key to understanding the GHS melting event is the presence of Eocene eclogites (some coesite-bearing) within the GHS, and broadly contemporaneous potassic lavas in southern Tibet. Comparison with present-day analogues of collisional orogenesis and the orogen-scale architecture of ancient ultrahigh-pressure terranes permits a simple model for burial, exhumation and partial melting of the GHS slab and generation of high-K magmas. Late Paleocene to early Eocene continental subduction to depths of around 200-300 km was terminated by decoupling (slab break-off) of the oceanic lithosphere in the mid Eocene. Partial melting of the leading edge of the slab, due to asthenospheric upwelling generated high-K magmas. The GHS quartzo-feldspathic wedge of recrystallized supracrustal material was propelled to the surface by buoyancy-driven extrusion, and was subject to fluid-absent decompression melting at ~35 km in the early Miocene.