Root system architecture (RSA) is important for rootstock adaptation to soil resource-limited environments, but our knowledge of root traits is hampered by the lack of high-throughput root phenotyping systems for perennial fruit crops such as grapevine. Numerous approaches have been developed for non-destructive observations of RSA with the support of advanced imaging and image analysis techniques. The most commonly used are those based on the periodic observation of roots in the soil through a transparent material. The objectives of the present study were (i) to develop a phenotyping system using rhizotrons for in situ quantification of root morphology without affecting shoot development, and then (ii) as an example of its application, to examine genetic intrinsic variations in root traits among two grapevine rootstock genotypes submitted to low soil water potential induced by the osmolyte, polyethylene glycol. Several root morphology traits (e.g., total root length, average root diameter, projected and surface area, plus length and area measurements for root diameter classes) were determined twice a week for a month using an A3 flat-bed scanner for 2D image acquisition followed by digital image analysis in WinRhizo software. Preliminary results demonstrated that the rhizotron phenotyping system was efficient in screening RSA for young grape rootstocks, making it very attractive in the future for routine genetics studies and other screening purposes.