We evaluate links between climate and simulated river bank erosion for one of the world's largest rivers, the Mekong. We employ a process-based model to reconstruct multidecadal time series of bank erosion at study sites within the Mekong's two main hydrological response zones, defining a new parameter, accumulated excess runoff (AER), pertinent to bank erosion. We employ a hydrological model to isolate how snowmelt, tropical storms and monsoon precipitation each contribute to AER and thus modeled bank erosion. Our results show that melt (23.9% at the upstream study site, declining to 11.1% downstream) and tropical cyclones (17.5% and 26.4% at the upstream and downstream sites, respectively) both force significant fractions of bank erosion on the Mekong. We also show (i) small, but significant, declines in AER and hence assumed bank erosion during the 20th century, and; (ii) that significant correlations exist between AER and the Indian Ocean Dipole (IOD) and El Niño Southern Oscillation (ENSO). Of these modes of climate variability, we find that IOD events exert a greater control on simulated bank erosion than ENSO events; but the influences of both ENSO and IOD when averaged over several decades are found to be relatively weak. However, importantly, relationships between ENSO, IOD, and AER and hence inferred river bank erosion are not time invariant. Specifically, we show that there is an intense and prolonged epoch of strong coherence between ENSO and AER from the early 1980s to present, such that in recent decades derived Mekong River bank erosion has been more strongly affected by ENSO.