A retrospective study was initiated to investigate the influence of recipients' Lewis subtype and HLA-matching on cadaveric kidney graft outcome. A total of 1111 patients receiving a first cadaveric kidney graft were analyzed. No difference in one-year graft survival was found between Lewis-negative (73%, n = 133) and Lewis-positive (73%, n = 978) recipients. Further subdivision of the study group into HLA-A,-B well-matched (0 or 1 mismatches [MM]) and poorly matched (2, 3, or 4 MM) revealed a strong deleterious effect of HLA-A,-B mismatching in the Lewis-negative group only. One year graft survival in Lewis-negative HLA-A,-B poorly matched (2, 3, or 4 A,B MM) patients was 60% (n = 67) versus 86% (n = 66) in the Lewis-negative HLA-A,-B well-matched (0 or 1 A,B MM) group (P = 0.004). For the Lewis-positive group the one-year graft survival rates were 72% (2, 3, or 4 A,B MM; n = 498) and 74% (0 or 1 A,B MM; n = 480), respectively (P = n.s.). The additional beneficial effect of HLA-DR matching again turned out to be strongest in the Lewis-negative group. In Lewis-negative, HLA-DR (0 MM) and -A,-B well-matched recipients (n = 36) graft survival was 94% versus only 64% in the Lewis-negative, DR matched, A,-B mismatched (2, 3, or 4 A,B MM) group (n = 25; P = 0.09). In the Lewis-positive, HLA-DR 0 mismatched group the one-year survival rates were 78% (0 or 1 A,B MM; n = 240) and 73% (2, 3, or 4 A,B MM; n = 253), respectively (P = 0.05). Our data suggest that donor recipient selection should not be based on Lewis matching per se. However, since Lewis-negative patients are at high risk of graft failure when receiving HLA mismatched kidneys, they should preferentially receive optimally HLA matched grafts.