Since the pioneering work of Patel and Terasaki, the presence of an anti-donor antibody of the IgG isotype, as demonstrated by a lymphocytotoxic assay on T cells, has become a contraindication to transplantation, due to the very high rate of graft loss reported (80% in the first few weeks post-transplant). The advent of more sensible and specific techniques of detection of anti-HLA antibodies (such as ELISA or Luminex techniques) has questioned this dogma, with a number of reports showing that transplantation, despite the presence of an donor-specific antibody (DSA), could be done without excessive graft losses, despite higher rates of rejection. We thus decided to retrospectively screen a cohort of patients consecutively transplanted in our unit. We then could determine the influence of preformed DSA, identified by HLA-specific ELISA or Luminex assays, on graft survival and evaluate the incidence of antibody-mediated rejection (AMR). Kidney graft survival at 8 years was significantly worse in patients with DSA. The incidence of AMR in patients with DSA was 9-fold higher than in patients without DSA and led to a significantly worse graft survival. The prevalence for AMR in patients with DSA detected on historic serum was 32.3% and was significantly more elevated in patients with strongly positive DSA and in patients with historic positive cytotoxic crossmatches. Interestingly, those patients with DSA detected by Luminex that did not experience AMR had worse graft survival that patients without DSA, supporting the concept of sub-clinical or chronic antibody-mediated rejection. As of today, the presence of preformed DSA is strongly associated with increased graft loss in kidney transplants, related to an increased risk of AMR. Our findings demonstrate the importance of detection and characterization of DSA before transplantation. Stratification of this immunological risk should be used both to determine kidney allocation and to elaborate specific strategies for these patients. In this regard, the avent of virtual PRA and virtual crossmatch may be used to assess transplant access for a given patient. We have thus devised an algorithm to maximize access to transplant of highly sensitized patients while keeping the immunological risk to a minimum.