In the “genetical genomics” fi eld, different approaches are proposed to improve characterization of QTL regions for various traits combining QTL detection and transcriptome profi ling. The most common approach is to identify genes whose eQTL colocalize with QTL of interest, providing new functional hypothesis about the QTL causative mutation. A second approach used only by Schadt et al (2003) and Le Mignon et al (2009) consists in performing linkage analysis for the trait using only some of the animal subgroups of a F2 population generated on the basis of their transcriptome profi les. Such an approach can refi ne some QTL and reveal other ones. This approach was applied to hepatic transcriptome profi les for 45 offspring of a chicken known to be heterozygous for a QTL for abdominal fatness (AF) on GGA5 at 168cM. 688 gene expressions signifi cantly correlated to the AF trait were obtained using a recent method taking into account the gene dependence independently of the trait of interest (Friguet et al, 2009; see ISAG2010 Blum et al). A hierarchical cluster analysis using these 688 genes distinguished fi ve groups among the 45 birds. After removing a subgroup of 7 animals, linkage analysis revealed another QTL on the GGA5 at ~102cM. These 7 animals presented the same paternal haplotypes, suggesting that the two QTL are in interaction. We show by different approaches (ANOVA, linkage analysis) a signifi cant interaction between the two QTL. These results show the power of the approach: transcriptome data allows separating a population into genetically homogenous subgroups, revealing the complexity of the genetic component of the complex traits.