Soil erosion is still severe in the Loess Plateau, and vegetation restoration practices are effective measures to fight it by reducing nutrient loss and improving soil quality. Soil microorganisms are important for soil ecosystem, which play crucial roles in biogeochemical cycling and ecosystem functioning. Soil biological and biochemical properties, such as microbial biomass, microbial community composition, and microbial functional diversity are frequently used for detecting accurate changes in soil properties. The major purpose of this study therefore, is to discover the effects of different vegetation restoration patterns on soil microbial community structures in the Loess Plateau. The study area was located in Yangjuangou catchment, a typical loess hilly area near Yan'an city of northern Shaanxi province, China. Four typical hillslopes, including F (artificial forest), G (grass), G-F-G (grass on the upper and lower slopes and artificial forest on the middle slope), and F-G-F (artificial forest on the upper and lower slopes and grass on the middle slope), were chosen. Several soil sampling sites, each covered by an area of 200m², were conducted on the summit, shoulder, backslope, footslope and toeslope along each hillslope, respectively. Three experimental plots in each site were conducted with an area of 25 m². Soil samples in the first two vertical layers (0-10 cm and 10-20 cm) were collected in August of 2007, for analyzing soil microbial community structure and soil physicochemical properties. Soil microbial community structure was analyzed by using phospholipid fatty acid (PLFA) analytical method and soil physicochemical properties were done by general soil and agricultural chemistry analytical methods. Results showed that significant difference of phospholipid fatty acid composition existed between different soil layers and soils with different vegetation restoration patterns, mainly in surface soil layer collected from G-F-G and F-G-F differing with that from F and G. The proportion of phospholipid fatty acid indicated fungi was higher in the surface soils under G-F-G and F-G-F, compared to F and G. Principal components analysis (PCA) showed that in the 0-10 cm soil layer, the phospholipid fatty acid structure of G-F-G was different with that of F and G, while the phospholipid fatty acid structure of F-G-F was different with F significantly. The biomass of soil bacteria, fungi, Gram-negative bacteria and Gram-positive bacteria under F-G-F pattern were all captured the highest values. There was significant difference of soil bacteria biomass and Gram-positive bacteria biomass in the surface soil layers among the four mentioned vegetation restoration patterns. The biomass of soil bacteria and fungi, however, had non-significant difference among 10-20 cm soil layers with different vegetation restoration patterns. Redundancy analysis (RDA) showed that soil organic carbon and total nitrogen had more important relationship with soil microbial community. However, the effects of some physicochemical properties such like pH and bulk density on soil microbial community needed more study. In conclusion, microbial community structure was dissimilar in soils and significant difference mainly appeared in surface soil layers under four vegetation restoration patterns. The roles of artificial forestation and forest-grass collocation on the slope in improving soil microbial biomass were believed much better than abandoned farmland.