A multivariate QTL detection was carried out on fatness and carcass composition traits on porcinechromosome 7 (SSC7). Single-trait QTLs have already been detected in the SLA region, andmultivariate approaches have been used to exploit the correlations between the traits to obtainmore information on their pattern: almost 500 measurements were recorded for backfat thickness(BFT1, BFT2), backfat weight (BFW) and leaf fat weight (LFW) but only about half that numberfor intramuscular fat content (IMF), affecting the detection. First, groups of traits were selectedusing a backward selection procedure : traits were selected based on their contribution to the linear combination of traits discriminating the putative QTL haplotypes. Three groups of traits could bedistinguished based on successive discriminant analyses: external fat (BFT1, BFT2), internal fat (LFW, IMF) and BFW. At least four regions were distinguished, preferentially affecting one or theother group, with the SLA region always influencing all the traits. Meishan alleles decreased all trait values except IMF, confirming an opportunity for marker-assisted selection to improve meat qualitywith maintenance of carcass composition based on Meishan alleles.