The thermal structure of the Tibetan plateau—the largest orogenic system on Earth—remains largely unknown. Numerous avenues provide fragmentary pressure/temperature information, both at the present (predominantly informed though geophysical observation) and on the evolution of the thermal structure over the recent past (combining petrological, geochemical, and geophysical observables). However, these individual constraints have proven hard to reconcile with each other. Here, we show that models for the simple underthrusting of India beneath southern Tibet are capable of matching all available constraints on its thermal structure, both at the present day and since the Miocene. Many parameters in such models remain poorly constrained, and we explore the various trade‐offs among the competing influences these parameters may have. However, three consistent features to such models emerge: (i) that present‐day geophysical observations require the presence of relatively cold underthrust Indian lithosphere beneath southern Tibet; (ii) that geochemical constraints require the removal of Indian mantle from beneath southern Tibet at some point during the early Miocene, although the mechanism of this removal, and whether it includes the removal of any crustal material, is not constrained by our models; and (iii) that the combination of the southern extent of Miocene mantle‐derived magmatism and the present‐day geophysical structure and earthquake distribution of southern Tibet require that the time‐averaged rate of underthrusting of India relative to central Tibet since the middle Miocene has been faster than it is at present.