Identifying phosphorus (P) sources is critical for solving eutrophication and controlling P in aquatic environments. Phosphate oxygen isotopes (δ 18 O p ) have been used to trace P sources. However, the application of this method has been greatly restricted due to δ 18 O P values from the potential source having wide and overlapping ranges. In this research, P sources were traced by combining δ 18 O p with multiple stable isotopes of nitrogen (δ 15 N), hydrogen (δD), and dissolved inorganic carbon (δ 13 C). Then, a Bayesian-based Stable Isotope Analysis in R (SIAR) model and IsoSource model were used to estimate the proportional contributions of the potential sources in the Tuojiang River. δ 18 O p was not in equilibrium with ambient water, and statistically significant differences in the δ 18 O p values were found between the potential sources, indicating that δ 18 O p can be used to trace the P sources. δ 15 N, δD, and δ 13 C could assist δ 18 O p in identifying the main sources of P. The SIAR and IsoSource models suggested that industrial and domestic sewage was the largest contributor, followed by phosphate rock and phosphogypsum and agricultural sewage. The uncertainty of the calculation results of the SIAR model was lower than that of the IsoSource model. These findings provide new insights into tracing P sources using multiple stable isotopes in watersheds.