The use of rooted walnut (Juglans regia L.) cultivars in orchards can result in an important lack of soil adaptation. Finding a genotype good enough to serve as both a cultivar and a rootstock seems quite improbable. Breeding programmes for cultivars result in the accumulation of genes for production and for leaf and/or stem diseases, resistance or tolerance. In these programmes, genotypes are not usually considered for soil adaptation, and it is known that clonal material behaviour is homogeneous both in favourable and in adverse soils. But when seedlings are used, a segregation of phenotypes for soil adaptation can be seen. This actually occurred when seedlings of cv. Serr, sown in 1998, were grown at the Instituto Murciano de Investigacion y Desarrollo Agroalimentario (IMIDA) on a sandy soil plot. Seedlings affected (or not) by Agrobacterium tumefaciens were found. Vigour (high, medium, or low), root mass (abundant, medium or weak), and root morphology (axial, with thick or with root hairs) were evaluated during the winters of 2000, 2001 and 2002 in plants damaged (or not) by A. tumefaciens. For walnut orchards, several J. regia rootstock seedlings have been selected to generate: a) tolerance to A. tumefaciens, b) rootstocks of diverse vigour, and c) root systems with abundant root mass. These seedlings could be used for clonal rootstock propagation and/or evaluated as mother plants for seed production.