Abstract: The homogenization of agricultural landscape caused by excessive agricultural intensification has been one of the main reasons for the reduction of farmland biodiversity. Research on the impact of the agricultural landscape pattern on the qualitative food web structure of pest-predatory natural enemies was conducted in Changtu County, Liaoning Province, where eight typical fields were sampled. Regression analysis and optimal model determined the relationship between food web parameters and landscape indexes. The results showed an insignificant correlation between food web interaction richness (IR) and landscape indexes. However, a significant positive multiple correlations between food web linkage density (LD) and contagion index (CONTAG, x₁) and aggregation index (AI, x₂) were observed. The corresponding optimal model was: LD=-64.621+0.780x₁+0.739x₂. The complexity of the qualitative food web structure of the corn pest-predatory natural enemies was dependent on the degree of concentration of the non-cultivated patches in the agricultural landscape. Furthermore, food web connectance (C) was significantly positively correlated with CONTAG, (x₁) and Shannon diversity index (SHDI, x₃), but was significantly negatively correlated with Shannon evenness index (SHEI, x₄). The corresponding optimal model was: C=-178.500+1.831x₁-106.808x₄; the more diverse the landscape types, the better the connectivity of the same patches; the more frequent the interaction between pests and predatory natural enemies, the more beneficial it is to maintain the complex food web structure. Food web generality (G) was significantly positively correlated with landscape shape index (LSI, x₅), cohesion index (COHESION, x₇), and AI (x₂); however, it was significantly negatively correlated with patch density (PD, x₆). The corresponding optimal model was: G=-2 994.798+26.891x₂+27.090x₅-0.491x₆+2.851x₇; the lower the degree of non-cultivated patch fragmentation, the stronger the search and aggregation behavior of natural enemies, which is beneficial and increases the stability of the food web structure. Food web vulnerability (V) was significantly positively correlated with the SHEI (x₄), but was significantly negatively correlated with CONTAG (x₁). The corresponding optimal model was: V=8.411+5.351x₄; the more evenly distributed patch types in the landscape, the higher the pest diversity and the increased complexity of the community structure. In general, the construction of the qualitative food web of corn pest-predatory natural enemies and the enhancement of anti-interference largely depends on the strength of the heterogeneity of the agricultural landscape. The use of field data in the construction of a food web matrix is a method that can be a powerful resource for studying ways to enhance the heterogeneity of agricultural landscapes.