Study region: Kaxgar, Yarkant, Hotan, Keriya, and Qarqan river basins in central Asia. Study focus: Precipitation is a crucial water source for sustaining arid regions in central Asia. However, due to sparse and unevenly distributed meteorological observations in mountain-basin systems, the spatial patterns and altitudinal gradient of precipitation remain unclear in central Asia. Based on recent observations from over 100 automatic weather stations spanning the Kunlun Mountains to the Tarim Basin, we evaluated 12 precipitation products and employed data fusion methods to generate a new dataset. New hydrological insights for the region: The application of a Feedforward Neural Network-based data fusion method effectively improves precipitation estimation in the mountain-basin systems of central Asia. Using the fused product spanning from 1979 to 2020, we updated the regional precipitation map and quantified the precipitation-altitude gradient for each river basin. The annual precipitation generally decreases from west to east, with mean values of 175.7 mm (Kaxgar), 126.8 mm (Yarkant), 106.7 mm (Hotan), 103.4 mm (Keriya), and 85.6 mm (Qarqan). During the past four decades, annual precipitation increased by 14.5, 10.3, 8.1, 5.6, and 3.3 mm per decade in the five river basins, respectively. The short-term measurements of automatic weather stations are important in data evaluation and improvement. In contrast, exclusive reliance on unevenly distributed national meteorological stations introduces substantial spatial biases in data evaluation.