Abstract: Lucerne (Medicago sativa) is widely planted in the Loess Plateau of western China and can accumulate soil carbon and nitrogen nutrients. However, continuous cropping of lucerne has consumed soil water and phosphorus for many years leading to a decrease in soil quality and alfalfa productivity. Therefore, after lucerne is planted for a certain period, it is necessary to plant the stubble and rotate annual crops to promote sustainable land use. Choosing suitable crops can improve the stability of the soil ecosystem. In this study, we analyzed the effects of long-term continuous cropping of alfalfa and rotation with annual crops on the structure and diversity of soil fungal communities in semi-arid areas based on a long-term localization experiment in the rainfed agricultural area of the Loess Plateau using the FUNGuild platform to predict the ecological functions of fungi in different treatments. The cropping systems included monocropping for 16 years of lucerne (LC), field fallow for 7 years after monocropping for 9 years of lucerne (rotation of lucerne-fallow, LF), field fallow for 2 years after monocropping for 9 years of lucerne and then planting corn (Zea mays) for 5 years (rotation of lucerne-fallow-corn, L_FC), planting potato (Solanum tuberosum) for 7 years after monocropping for 9 years of lucerne (rotation of lucerne-potato, LP), and planting millet (Panicum miliaceum) for 7 years after monocropping for 9 years of lucerne (rotation of lucerne-millet, LMi). A total of 7 phyla, 25 classes, 77 orders, 169 families, and 347 genera of fungi were identified. The fungi were mainly Ascomycota, Zygomycota, and Basidiomycota at the phylum level. Ascomycota was the first dominant phylum in different treatments, and its relative abundance ranged from 69.17% to 88.22%, which was much greater than that of the subdominant phyla Zygomycota (6.72%–19.88%) and Basidiomycota (1.64%–9.01%). The dominant genera varied in different treatments, with Phaeomycocentrospora in LC treatment, Gibberella in LF, LP, and LMi treatments, and Mortierella in L_FC treatment. Redundancy analysis revealed that soil available phosphorus (P=0.002) was the main factor influencing the soil fungal community structure. The alpha diversity results showed that crop type had no significant impact on the diversity and richness of the soil fungal communities. However, the Shannon index was significantly negatively correlated with the soil nitrate-nitrogen content (r=−0.553, P<0.05) and the Simpson index was significantly positively correlated with the soil nitrate-nitrogen content (r=0.515, P<0.05). Functional prediction with FUNGuild showed that pathotrophs (25.44%–39.27%) was the dominant fungal functional group of loessal soil in this study. After lucerne rotation with annual crops, the relative abundance of transitional fungal groups, such as soil saprotrophs-symbiotrophs, pathogens-saprotrophs-symbiotrophs, and pathotrophs-saprotrophs-symbiotrophs, changed, whereas the relative abundance of pathotrophs-saprotrophs decreased. Available phosphorus (P=0.002) and nitrate-nitrogen (P=0.02) were the main environmental factors affecting the changes in soil fungal functional groups. In conclusion, rational cropping systems are conducive in enriching the structure of soil microbial communities and promoting the stability of the soil ecosystems in the region. The results of this study can provide reference and data support for the prediction of soil fungal communities and their functions in different planting systems.