In order to develop management plans for water and soil conservation, it is necessary to determine the sources of sediment production in watersheds. During the past three decades fingerprinting technique has been used extensively in determining the contribution of different sources of sediment. In this study, was carried out sediments fingerprinting and determine the contribution of each source to sediment production of the Toulbane watershed in Golestan province. To this end, 44 source samples were collected from forest, pasture, agriculture and bank erosion. Also 8 sediment samples were collected using Philips time-integrated sediment sampler. Afterward, the concentration of 34 geochemical properties was examined in the laboratory using the ICP device. Next, the optimal composite tracers were determined to discriminate sediment sources by using statistical tests including mass conservation test and Kruskal-Wallis. The contribution of different sources to sediment production was determined using the multivariate mixing model. Finally, the uncertainty in the case of a low number of data, was examined using the Monte Carlo method. As a result, after statistical tests, 12 tracers were selected as the optimal composite fingerprints. The bank erosion was main source to sediment production with 52.18% and the forest had the lowest contribution to sediment production with 4.39%. The contribution of agriculture and pasture was 33.23% and 10.21%, respectively. According to the uncertainty analysis, bank erosion is the most significant source to sediment production. Also, the high difference between the upper and the lower boundaries in different sources indicates high uncertainty.