Root system architecture (RSA) plays a key role in the adaptive response of crops to drought conditions. Due to the difficulty in phenotyping roots under field conditions, the genetic control of root features has often been investigated in plants grown under controlled conditions. Along this line, a panel of 188 durum wheat (Triticum durum Desf.) elite accessions have been evaluated for RSA traits at the seedling stage in order to identify the relevant QTLs and compare their position with that of QTLs obtained for grain yield (GY) in a previous study where the same accessions were evaluated in 15 field trials characterized by a broad range of soil-moisture availability and GY (Maccaferri et al. 2011, J Exp Bot, 62: 409-438). In total, nine RSA features were measured in the seminal and primary roots. Highly significant differences among genotypes were detected for all traits except for root hairiness. For this study, the 15 field trials reported in Maccaferri et al. (2011) were grouped in low-yielding environments (LYE; average GY of 3.09 Mg ha^-1) and high-yielding environments (HYE; GY of 6.81 Mg ha^-1). Two significant, albeit low, correlations between GY and RSA traits were identified exclusively across LYE. In particular, GY was positively correlated with number of roots per plants (RN) and total roots length (r = 0.26 and 0.21, respectively). The accessions in the upper 20^th percentile for RN (5.8 roots per plant) outyielded by 4.5% the accessions in the lower 20^th percentile (4.4 roots per plant). Association mapping analysis of RSA traits is underway based upon ca. 9,000 SSR, DaRT and SNP markers. The position of the GY QTLs reported by Maccaferri et al. (2011) will be compared with the position of the QTLs for RSA features identified in this study.