Understanding the trade-offs and synergies of ecosystem services (ESs) is of great significance for guiding sustainable management practices within ecosystems. Based on multi-source data on land use and land cover, meteorology, soil types, normalized difference vegetation index (NDVI), and socio-economic statistics, the spatio-temporal variation of five ESs, namely, food supply, carbon sequestration, soil conservation, sand fixation, and water yield, was analyzed in Duolun County, the agro-pastoral ecotone of Northern China. We also identified hot spots for these five ESs' supplies. In addition, based on the correlation analysis method, the trade-offs and synergies between ESs were analyzed at three scales: county, townships, and different land use types. Results showed that the food supply and soil conservation increased significantly, whereas the water yield and carbon sequestration increased slightly, the sand fixation had an evident decrease, and the spatial pattern of the five ESs basically remained unchanged from 2000 to 2015. The area of the first type hot spot in Duolun county clearly increased, while the area of the third type hot spot decreased. Grasslands with a proportion of more than 50% of the county area could provide these five ESs better than the cultivated land. Trade-off relationships in Duolun County mainly existed among three pairs of ESs: water yield and soil conservation, water yield and carbon sequestration, and water yield and food supply. Due to the spatial and environmental heterogeneity of the study area, there are four pairs of ESs at the township scale were consistent with those at the county scale, and three pairs of ESs at different land use types were consistent with those at the county scale. The combination of different land use types, elevation, soil types, and other factors made the trade-offs/synergies between townships and county in Duolun County obviously different. Therefore, at the agro-pastoral ecotone, where the ecological environment is highly complex, it is more necessary to select the appropriate spatial scale for the trade-offs/synergies relationship analysis and ecosystem management than in other, less complex, ecosystems.