The influence of mismatches for HLA-DR ''split'' specificities was investigated in more than 8000 cadaver kidney transplants. HLA-DR typing was performed using DNA methodology. Among first transplants, mismatches defined by HLA-DR split specificities did not have a deleterious influence. Among retransplants, however, graft survival was significantly decreased if a mismatch was defined, considering split specificities in patients with no mismatch according to the ''broad'' definition (P=0.04) and also in grafts with two split mismatches, which showed only one mismatch according to the broad definition (P=0.03). Moreover, consideration of further ''subsplit'' specificities resulted in clinically relevant mismatches only among retransplants. These data indicate that the recognition of HLA-DR split specificity mismatches is fundamentally different in primary and regraft recipients, The results imply that recipients and donors of kidney retransplants should be typed for HLA-DR split specificities and that these specificities should be considered for organ allocation.