The effectiveness of five selection methods for genetic improvement of net merit comprising trait 1 of low heritability (h2 = 0.1) and trait 2 of high heritability (h2 = 0.4) was examined: (i) two-trait quantitative trait loci (QTL)-assisted selection; (ii) partial QTL-assisted selection based on trait 1; (iii) partial QTL-assisted selection based on trait 2; (iv) QTL-only selection; and (v) conventional selection index without QTL information. These selection methods were compared under 72 scenarios with different combinations of the relative economic weights, the genetic correlations between traits, the ratio of QTL variance to total genetic variance of the trait, and the ratio of genetic variances between traits. The results suggest that the detection of QTL for multiple-trait QTL-assisted selection is more important when the index traits are negatively correlated than when they are positively correlated. In contrast to literature reports that single-trait marker-assisted selection (MAS) is the most efficient for low heritability traits, this study found that the identified QTL of the low heritability trait contributed negligibly to total response in net merit. This is because multiple-trait QTL-assisted selection is designed to maximize total net merit rather than the genetic response of the individual index trait as in the case of single-trait MAS. Therefore, it is not economical to identify the QTL of the low heritability traits for the improvement of total net merit. The efficient, cost-effective selection strategy is to identify the QTL of the moderate or high heritability traits of the QTL-assisted selection index to facilitate total economic returns. Detection of the QTL of the low h2 traits for the QTL-assisted index selection is justified when the low h2 traits have high negative genetic correlation with the other index traits and/or when both economic weights and genetic variances of the low h2 traits are larger as compared to the other index traits of higher h2. This study deals with theoretical efficiency of QTL-assisted selection, but the same principle applies to SNP-based genomic selection when the proportion of the genetic variance ‘explained by the identified QTLs’ in this study is replaced by ‘explained by SNPs’.