In order to understand the effect of soil fertility under different rotation patterns of "grain, cash, and forage crops" in the semi-arid area of northwest China, seven rotation patterns, including corn-pea (C-Pe), corn-corn (C-C), 2-year alfalfa (2A), broomcorn-potato (B-Po), oat-corn (O-C), potato-oat (Po-O), pea-broomcorn (Pe-B), were set up in Yuanzhou District, Guyuan City, Ningxia Autonomous Region in 2017. The Indicators including the soil nutrient, enzyme activity and characteristics of culturable microbial community under different grain rotation cropping patterns were measured. Meanwhile, the soil fertility was comprehensively evaluated, which provide theoretical basis and practical guidance for local farming improvement. The results showed:(1) Different rotation patterns had significant effects on soil chemical properties. Compared with other rotation models, 2A was more conducive to the accumulation of soil nutrients. However, C-C mode increased the consumption of nutrients. The organic matter, total nitrogen, available nitrogen, available phosphorus, and available potassium content of C-Pe and Pe-B modes were significant higher than those of B-Po and O-C modes, moreover, rotation treatment had less effect on total potassium content. (2) Different crop rotation patterns had significant effects on soil enzyme activities. The soil urease and sucrase activities of 2A mode were significantly higher than those of other rotation patterns (P<0.05). The soil enzyme activity in C-C mode was the lowest, and the sucrase, the urease, and the alkaline phosphatase activities of O-C and Pe-B modes were significant higher than those of B-Po mode. (3) Different crop rotation patterns had significant effects on the number of soil culturable microorganisms. In crop rotation patterns, the number of soil microorganisms was dominated by bacteria, followed by actinomycetes and fungi with obvious vertical distribution of soil layers. Compared with other rotation models, the number of bacteria and the total number of microorganisms in 2A, C-C, and B-Po models decreased significantly and the number of fungi increased significantly (P<0.05), which was opposite to O-C mode. 4) Principal component analysis-numerical clustering method was used to comprehensively evaluate the soil fertility of 15 indicators of soil nutrient, soil enzyme activity, and soil microbial population under different rotation patterns. The results showed that the cumulative contribution rate of the two principal components extracted by principal component analysis was up to 89.38%. The first main component was composed of soil organic matter, total nitrogen, available nitrogen, total phosphorus, available phosphorus, available potassium, pH, sucrase, urease, and alkaline phosphatase, and the cumulative contribution rate was up to 63.56%. The second main component was composed of bacteria, fungi and microorganisms and cumulative contribution rate reached 25.82%. The overall score of each rotation pattern on the two principal components was ranked as 2A > Pe-B > C-Pe > O-C > Po-O > B-Po > C-C. The two principal component scores were further clustered, and seven models were divided into three categories:the first type (2A) had the best soil fertility; the second type (including Pe-B, O-C, C-Pe, Po-O, and B-Po modes) had better soil fertility; and the third type (C-C) had poor soil fertility. These research results may provide theoretical basis and practical guidance for the improvement of local farming methods.