A growing awareness of the wider environmental significance of diffuse sediment pollution in interconnected river-lake systems has generated the need for reliable provenance information. Owing to their insufficient ability to distinguish between multiple sources, common sediment source apportionment methods would rarely be a practical solution. On the basis of the inseparable relationships between sediment and adsorbed microorganisms, community-based microbial source tracking may be a novel method of identifying dominant sediment sources in the era of high-throughput sequencing. Dongting Lake was selected as a study area as it receives considerable sediment import from its inflowing rivers during the flood season. This study was conducted to characterize the bacterial community composition of sediment samples from the inflow-river estuaries and quantify their sediment microbe contributions to the central lake. Metagenomic analysis revealed that the community compositions of source sediment samples were significantly different, allowing specific sources to be identified with the machine learning classification program SourceTracker. A modified analysis using SourceTracker found that the major contributors to three major lake districts were the Songzi, Zishui, and Xinqiang Rivers. The impacts of hydrodynamic conditions on source apportionment were further verified and suggested the practicability of this method to offer a systematic and comprehensive understanding of sediment sources, pathways, and transport dynamics. Finally, a novel framework for sediment source-tracking was established to develop effective sediment management and control strategies in river-lake systems.