A common feature of convergent plate boundaries is the self-organization of strain, exhumation and topography along discrete, arcuate boundaries. Deviations from this geometry can represent first-order changes in stress applied at a plate boundary that must affect how strain is partitioned within the interior of an orogen. The simplicity of the Himalayan fold and thrust belt seen along its central portion breaks down along the eastern extremity of the arc where the 400 km-long Shillong Plateau has developed. This change in strain partitioning affects nearly 25% of the arc and has not previously been considered to be important to the orogen's development. New low-temperature thermochronometry data suggest this structure initiated in mid to late Miocene time, significantly earlier than was previously estimated from the sedimentary record alone. Development of the Shillong Plateau may be linked to a number of kinematic changes within the Himalayan and Burman collision zones that occur at the same time. These events include the onset of E–W extension in central Tibet, eastward expansion of high topography of the Tibetan Plateau, onset of rotation of crustal fragments in southeastern Tibet, and re-establishment of eastward subduction beneath the Indo-Burman ranges. We suggest that the coincidence of these tectonic events is related to the ‘dismemberment’ of the eastern Himalayan arc, signifying a change in regional stress applied along the India–Eurasia–Burma plate boundaries. Discrepancies between vertical long-term faulting rates and geodetically derived far-field convergence rates suggest that the collisional boundary in the eastern Himalayan system may be poorly coupled due to introduction of oceanic and transitional crust into the eastern plate boundary. The introduction of dense material into the plate boundary late in the orogen's history may explain regional changes in the strain field that affect not only the Himalaya, but also the deformation field more than 1000 km into the Tibetan Plateau.