Abstract: The agricultural landscape with high proportion and diversity of non-cropped habitats is the key to maintaining agricultural biodiversity. It is a hypothesis that the natural enemy population and types of corn borer could increase by complex and high composition of non-cropped habitats in cornfields. To clarify the relationship between non-cropped habitats proportion and natural enemy, this study investigated the effects of composition of non-cropped habitats on natural enemy population of corn borer in Changtu County in the Northeast China Plain. Twenty plots (φ=600 m) were selected and classed into five groups based on the proportions of non-cropped habitats (0-10%, 10%-20%, 20%-30%, 30%-40% and 40%-50%). The samples of natural enemies of corn borer were collected by the pitfall trap method. The families of enemies were identified using stereomicroscope. Then principal component analysis was used to classify the main types of non-cropped habitats in the region. Stepwise regression model was used to fit relation curve of non-cropped habitat composition (ratio and type) and population of natural enemies of corn borer. Then the optimal model was selected based on the Akaike's Information Criterion (AIC) for effects of different proportions of non-cropped habitats on natural enemy population of corn borer. Results showed that natural enemy population of corn borer increased at first and then decreased with increasing proportion of non-cropped habitats. Natural enemy population of corn borer was highestwith the optimal model of AIC_{minimum value}=4.24, Y=b₀-b₁D + b₂PC2, PC2 (β=3.787, P=0.005) when the proportion of non-cropped habitats was 20%-30%. This optimal model implied that there was significant positive correlation between natural enemy population of corn borer and the composition of non-cropped habitats dominated by woodland. There was also significantly negative correlation between natural enemy population of corn borer and the landscape diversity index D. Other modelsPC1 (β=1.957, P=0.000) for 20%-30% proportion of non-cropped habitats implied that significant positive correlation existed between natural enemy population of corn borer and the composition of non-cropped habitats dominated by grassland and village. Under other proportions of non-cropped habitats, the optimal models suggested that corn filed neighboring woodland or other crops fields was beneficial for increase of natural enemy population of corn borer. In conclusion, the study showed that natural enemy population of corn borer was highest when the proportion of non-cropped habitats was 20%-30%. Non-cropped habitats dominated by woodland, grassland, village or in intercropped patterns attracted natural enemies of corn borer in farmlands with different proportions of non-cropped habitats. The study further provided theoretical basis for the prevention and control of corn borer by ways of landscape ecology.