A better understanding of phosphorus-transfer process and influence factors at Sediment-Water Interface (SWI) is essential to develop effective and efficient river managements strategies. In this study, overlying water, pore water and riverbed sediment samples were collected in the Huaihe River (HR), a highly polluted river in Eastern China, in May 2013, July 2013 and June 2016, respectively. Models were developed to analyze influence factors on phosphorus by Bayesian Networks (BNs), which could describe complex interdependencies between dependent/independent variables conveniently compared to traditional statistical models. The transfer process and effects of land use on phosphorus at the SWI were evaluated. Results indicated that phosphorus concentration decreased in overlying water and pore water similarly but increased in riverbed sediment from 2013 to 2016. The concentrations in overlying/pore water reached a maximum at the middle of the research area, while those in riverbed sediment increased from upstream to downstream. Phosphorus was transferred from overlying water to riverbed sediment through pore water mainly in a dissolved phase, indicating that riverbed sediment could absorb contaminants and act as “pollution sink” in the HR. Upstream water and sediment were important sources of downstream phosphorus with the influence coefficients ranging from 0.294 to 0.491. Land use within 1 km river buffer had most significant influence on total phosphorus in overlying water and 2 km river buffer to riverbed sediment while land use had the most noteworthy effects on particulate/dissolved phosphorus in 2 km and 8 km river buffer respectively. Farmland, urban and rural residential land were important sources of phosphorus at the SWI, and farmland contributed most (>30%) in the HR. The study not only provides insights into phosphorus-transfer process and influence factors at the SWI in the HR, but the proposed model also could be applicable in other polluted rivers.