Mechanism for fault segmentation in thrust belt is a key to understanding the orogenic process and seismic risks. A ~50 km long aftershock gap emerged between the ruptures of the 2008 Wenchuan and the 2013 Lushan earthquakes along the eastern margin of the Tibetan Plateau. Previous studies suggested that weak materials under ductile deformation cause the gap. Here we propose an alternative explanation: differential erosion drives the along-strike variation in fault activity. To testify the two competing models, we conducted low-temperature thermochronology and fluvial shear stress analyses to depict the spatial distributions of erosion. We obtained eight apatite fission track dates (6-44 Ma) in the gap and deduced erosion rates of 0.5-0.6 mm/yr and 0.3-0.4 mm/yr since ~8 Ma in the hanging -wall and footwall of the Shuangshi-Dachuan fault, respectively. We carried out linear fitting based on an empirical relationship between thermochronology-derived erosion rate and fluvial shear stress, and then calculated the erosion rate for each survey point of fluvial shear stress. Our new data reveal that in the hinterland, the erosion rate at the gap is lower than that of adjacent areas along strike, whereas in the range front, the erosion rate at the gap is greater. This spatial pattern supports the “differential erosion” hypothesis and is at odds with the “weak material” model. This study illustrates that heterogeneous erosion regulates fault segmentation in this thrust belt. Moreover, the aftershock gap acts as a barrier for the past major earthquakes, which poses substantial seismic potential to this region.