Previous analysis of triangulation data of the Survey of India concluded that the great 1897Assam earthquake occurred on a south dipping fault near the northern edge of the Shillong Plateau, which was named the Oldham fault. This attribution has been questioned on geological and geodetic grounds. We refine the triangulation data, adding recently discovered observations, and demonstrate that they require average slip of 25 ± 5 m on a fault that dips south at ∼40 ∘ beneath the plateau. The best fitting solution to the geodetic observations gives a rupture length of 79 km. However, the Chedrang fault, immediately to the west of the Oldham fault, appears to have slipped as a subvertical tear fault during or shortly after the 1897 earthquake, with over 10 m of down-to-the-west normal-sense slip. This observation suggests that the western end of the main rupture approached within a few kilometers of the Chedrang fault, giving a length of 95 km for the rupture. This range of parameters gives a magnitude 8.15< M w <8.35 for the earthquake. The triangulation data cannot be satisfied by slip on the north dipping faults that border the southern edge of the Shillong Plateau nor by slip on a south dipping fault that has been postulated in the Brahmaputra Valley. GPS velocities show that up to 5 mm/yr of shortening is taken up across the plateau and its borders; this suggests, via moment-frequency relations, that the interval between great earthquakes in the region is several thousand years but that earthquakes of magnitude 7 or greater should occur roughly once per century.