Fingerprinting techniques can be used to quantify sand source contributions with applied relevance, such as the management of sand accumulation problems for desert railways traversing complex aeolian landscapes. This study applies the FingerPro model to elucidate the sand sources impacting the Golmud-Korla Railway (GKR) in northwest China. Sand samples were collected from three distinct sources: the Gobi, a low-lying coarse sand sheet (LCSS), and a dune, alongside mixture samples from sand deposits on railway fences. Sand (< 100 mu m) deposited on railway sand fences predominantly originated from the LCSS, contributing 83.44% and 76.59% to the < 63 mu m and 63-100 mu m particle size fractions, respectively. Conversely, the dune was the dominant source for particles ranging in size from 100 to 500 mu m, accounting for 75.40%. Annually, an estimated 1388.0 m(3) of sand from LCSS and 4987.2 m(3) of sand from the dune are transported to the 2-km-long sand fence in the upwind direction of the GKR. In terms of per-unit-area contribution, LCSS contributes the most (more than 3,700 m(3).km(-2)a(-1)), followed by the dune (1,534 m(3).km(-2)a(-1)), while the Gobi contributes the least (29.38 m(3).km(-2)a(-1)). These findings underscore the need for integrated sand control measures that address both LCSS and dunes. Consolidation of LCSS is needed to suppress dust emissions that affect railway equipment. Targeted dune control is needed to manage hazards from larger sand-sized particles (100-500 mu m) that obstruct railway sand fences. For arid regions with complex sedimentary environments, we recommend using differentiating particle size ranges in source differentiation analyses to capture variations in sediment grain size distributions with greater precision.