Abstract: Abstract: Sediment control has attracted more attention in watershed management planning, especially for the transport of contaminants and nutrients (such as phosphorus (P)) by fine sediment. Implementation of specific management solutions for a watershed requires understanding the location and nature of the major sediment sources. Composite fingerprinting techniques have been proved as an effective mean of sediment source apportionment in other countries. However, research using this method is not reported for small watersheds that are relative homogeneity in underlying surface, or have lower sediment load output and more human disturbance. As for nutrients, particulate P adsorbed by fine sediment is the major part of P output in an agricultural watershed. Therefore, successful sediment source apportionment would give an access to P source identification and quantitative apportionment study. The objectives of this paper were to identify the sources of particulate P in a small catchment of the upper Jiuxiang River in Nanjing, and to estimate the relative contributions of the potential sources based on sediment source fingerprinting. A total of 56 surface samples from potential sediment sources were collected, including 20 from woodland, 21 from cultivated land, and 15 from mine areas and roads of transportation. In addition to suspended sediment collection, 17 deposited sediment samples were collected in the mainstream near catchment outlet. Twenty three potential fingerprint properties were selected for laboratory analysis with all samples screened to < 63 μm. Composite fingerprints were acquired by a two-stage statistical fingerprinting procedure and a multivariate mixing model was used to estimate the contributions of potential sediment sources. A goodness of fit model was used to test the performance of the multivariate mixing model. The particulate P source apportionment was then proceeded based on the sediment fingerprinting results and P content information from all potential sources. The apportionment results showed that cultivated land contributed 25.3%-65.2% of the sediment yield and 52.2%- 85.8% of the particulate P yield in the studied watershed. The mine and road area contributed 34.8%-74.7% of the sediment yields but only 14.2% - 47.7% of the particulate P yield. Less than 0.1% of the sediment or particulate P yield was from woodland. The results showed that cultivated land was a more importance as a particulate P source than as a sediment source due to the high P content of cultivated soils. While the mine and road areas were less important as a particulate P source for their lower P content. As for woodland, relative lower P content and long distance to receiving water made it less important as both sources. Although there still have some significant improvements in the future work, such as prior knowledge and selection of potential sources, optimization for fingerprinting procedure and the consideration of P transform in transport, etc., we concluded that particulate P source apportionment based on sediment fingerprinting could provide a new method to improve the particulate P source apportionment work in the watershed with limited observation data and would be helpful in diffuse source pollution control.