Abstract: The influence of land use/cover patterns on soil erosion and sediment transport can be one of the most important indicators to optimize the soil and water conservation measures in a watershed. However, it is still lacking in the quantitative evaluation of the land use/cover pattern and their impacts on soil erosion and sediment yield at the watershed scale. This study aims to explore the long-term changes in the sediment yield on the interannual and intermonthly temporal scales from 1982 to 2020, together with their responses to the land and use/cover pattern. A typical watershed was selected in the Loess Plateau, and the Lyuergou watershed. The M-K trend analysis was employed to elucidate the long-term temporal trends of annual rainfall, runoff, and sediment yield in the study area. An investigation was explored on the changes in the runoff and sediment under different vegetation patterns using the land use/cover pattern. The index of connectivity was chosen to characterize the impacts of the land use/cover pattern on soil erosion and sediment yield. Accordingly, quantitative relationships were also established between the index of connectivity and sediment yield. In addition, the monthly rainfall amount, the maximum daily rainfall amount, and erosive rainfall days were selected as the indices to classify the rainfall types. The results indicated that the Lyuergou watershed experienced cropland abandonment and vegetation restoration, where the index of connectivity decreased significantly during 1982-2020. Consequently, the whole study period was divided into four stages: 1982-1985, 1986-1990, 1991-2000, and 2001-2020, corresponding to the land use/cover pattern Ⅰ, Ⅱ, Ⅲ, and Ⅳ, respectively. The average annual sediment yield varied consistently, as the land use/cover pattern changed. Significantly downward trends were found in the average annual sediment yield and runoff depth, whereas, the annual rainfall amount showed no significant changing trend. Specifically, the annual runoff depth and sediment yield of land use/cover pattern Ⅳ decreased by 74.5% and 68.1%, respectively, compared with the land use/cover pattern Ⅰ, respectively. The average monthly runoff depth and sediment yield in the flood season were 11.7, 5.9, 1.8, 4.0 mm and 713.4, 377.4, 112.0, 317.9 t/km2 in the land use/cover pattern Ⅰ, Ⅱ, Ⅲ, and Ⅳ, respectively. The runoff depth and sediment yield under the later land use/cover pattern (Ⅲ and Ⅳ) were significantly lower as opposed to those in the early land use/cover pattern Ⅰ. The erosive rainfall in the flood season was classified into four types: type A (minor rainfall amount, and weaker rainfall erosivity), type B (medium rainfall amount, and weaker rainfall erosivity), type C (great rainfall amount, and medium rainfall erosivity), and type D (great rainfall amount, and larger rainfall erosivity). Type A, B and C contributed to the most sediment yield on the small watershed. However, there was no significant difference in the monthly runoff depth and sediment yield among different land use/cover patterns under type C. The sediment yield on both interannual and intermonthly scales increased exponentially with index of connectivity, whereas the relationships were weaker under type C. The finding can provide a strong reference to the optimal allocation of land use/cover patterns and soil and water conservation measures in arid and semi-arid areas.