Mapping the distribution of locked segments along subduction megathrusts is essential for improving quantitative assessments of seismic hazard. Previous geodetic studies suggest the Main Himalayan Thrust (MHT) is homogeneously locked (or coupled) along its complete length over a down-dip extent of ~100 km. However, an increasing number of seismological and geophysical observations suggests the MHT is structurally segmented along strike. Furthermore, coupling appears to vary significantly in subduction zones worldwide, hence the MHT appears as a striking anomaly. Here, we use a recent compilation of geodetic data and a fully Bayesian approach to show that coupling is highly heterogeneous along the MHT, thus reconciling all observations available. Our probabilistic estimate of interseismic coupling along the entire MHT highlights four large, highly-coupled patches separated by three potential barriers of low coupling. Locked patches overlap with estimated rupture areas of historical large earthquake over the past centuries. The coincident spatial variability in coupling, seismicity, prominent active topography, and complexity in earthquake history, suggests a structural segmentation of the Himalayan imposed by inherited tectonic structures from the India-Eurasia collision. This correlation implies that inherited tectonic structures may influence how stress builds up along the MHT and thus may influence the location and size of large Himalayan earthquakes and the growth of the Himalaya.