Microorganisms are key participants in regulating the multifunctionality of ecosystems, but the specific connection between soil microbial communities and ecosystem multifunctionality remains unclear, especially in restored ecosystems in sandy areas. To address this, this study focused on four vegetation types in the Mu Us Sandy Land, measuring soil physical and chemical properties, carbon components, microbial community composition, diversity, and enzyme activity. The single functions of soil organic carbon stabilization (SCS), nutrient cycling (NC), and microbial activity (MA) and ecosystem multifunctionality (EMF) were calculated using the average value method and the single domain value method to clarify the regulatory role of soil microbial diversity and community composition on soil functions under different vegetation types. The results showed that: (1) Among the four vegetation types, Acidobacteria, Proteobacteria, and Ascomycota were the dominant microbial groups; the Ace index and Chao1 index of bacteria and fungi showed no significant differences among the four vegetation types, while the Simpson index was significantly higher in typical grassland than in desert steppe and artificial forest. (2) The single function values of the ecosystem calculated by the average value method showed significant differences, with the highest values in typical grassland and the lowest in desertified grassland or artificial forest. The multifunction values were TS > DS > AF > GD in both methods. (3) The single function values in typical grassland were higher than those in other vegetation types, and SCS contributed the most to EMF. The contribution of NC to EMF in desert steppe was the highest (0.47) as calculated by the average value method. (4) Glomeromycota fungi, Firmicutes, and some uncategorized fungi were positively correlated with MA, SCS, NC, and EMF, while Actinobacteria and Ascomycota were negatively correlated with them. (5) Vegetation types mainly regulated EMF by influencing the composition of the fungal community. Overall, fungi had a higher explanatory power for EMF than bacteria. This study clarified the influence of microbial communities in different vegetation types in the Mu Us Sandy Land on EMF, providing data support and theoretical basis for local ecological restoration and ecosystem management.