Identification of sediment sources within catchments is crucial for river and catchment management. A multivariate mixing model in conjunction with two sets of statistically verified composite fingerprints was employed to identify the main sediment source areas at five locations along the main channel of the Shouchang River, southeastern China. The fingerprint composite fingerprints derived from the principle component analysis (PCA) and discriminant function analysis (DFA) combination can better apportion the five types of potential sediment sources. This combination demonstrated that over 33.0% sediment at each location came from river bed, with the largest contribution of 42.9%. The contributions from construction lands ranged from 16.0 to 26.8%. Sediment contributions from forest lands ranged from 7.8 to 22.1%. A large portion of sediment came from paddy fields, varying from 8.7 to 31.4% at the five locations. The sediment from dry land was the smallest, with the largest contribution of 18.4%, and the least contribution of 4%. Similar findings were also identified using the composite fingerprints derived from KW-H and DFA combination. Land use, topography, precipitation, and human activities were responsible for the sediment source apportionments and their variations at the five sediment sampling sites. This study implies that the available sediment from road and bridge built as well as plant production should well be supervised, laws should be made to inhibit in-stream sand mining activity, and the connectivity between paddy fields and channels should be reduced.