The efficiency of marker-assisted selection for native resistance to European corn borer stalk damage can be increased when progressing from a QTL-based towards a genome-wide approach. Marker-assisted selection (MAS) has been shown to be effective in improving resistance to the European corn borer (ECB) in maize. In this study, we investigated the performance of whole-genome-based selection, relative to selection based on individual quantitative trait loci (QTL), for resistance to ECB stalk damage in European elite maize. Three connected biparental populations, comprising 590 doubled haploid (DH) lines, were genotyped with high-density single nucleotide polymorphism markers and phenotyped under artificial and natural infestation in 2011. A subset of 195 DH lines was evaluated in the following year as lines per se and as testcrosses. Resistance was evaluated based on stalk damage ratings, the number of feeding tunnels in the stalk and tunnel length. We performed individual- and joint-population QTL analyses and compared the cross-validated predictive abilities of the QTL models with genomic best linear unbiased prediction (GBLUP). For all traits, the GBLUP model consistently outperformed the QTL model despite the detection of QTL with sizeable effects. For stalk damage rating, GBLUP's predictive ability exceeded at times 0.70. Model training based on DH line per se performance was efficient in predicting stalk breakage in testcrosses. We conclude that the efficiency of MAS for ECB stalk damage resistance can be increased considerably when progressing from a QTL-based towards a genome-wide approach. With the availability of native ECB resistance in elite European maize germplasm, our results open up avenues for the implementation of an integrated genome-based selection approach for the simultaneous improvement of yield, maturity and ECB resistance.