Genetically diverse populations can increase plant resistance to natural enemies. However, beneficial genotype pairs remain elusive due to the occurrence of positive or negative effects of mixed planting on plant resistance, respectively called associational resistance or susceptibility. Here, we identified key genotype pairs responsible for associational resistance to herbivory using genome-wide polymorphism data of the plant species Arabidopsis thaliana under field conditions. To quantify neighbor interactions among 199 genotypes grown in a randomized block design, we applied a genome-wide association method named “Neighbor GWAS” for genomic prediction. These analyses predicted that 823 of the 19,701 candidate pairs could reduce herbivory by mixed planting. To test this prediction, we planted three pairs with expected positive effects in mixtures and monocultures. Significant reductions in herbivore damage were detected for pairs with the largest and second-largest predicted effects, whereas the pair with the smallest predicted effects did not exhibit significant effects of the mixed planting treatment on herbivore damage. Our study illustrates the power of genomic prediction to assemble genotype mixtures with positive biodiversity effects on ecologically important traits.