We previously reported that delayed graft function (DGF) in the absence of biopsy-proven acute rejection (Rej) had no effect on outcome of primary cadaver kidney transplantation (TX). By contrast, DGF in combination with Rej strongly predicted poor long-term graft survival. We asked whether this poor long-term outcome was due to early graft loss associated with DGF, or to an ongoing process leading to late graft loss. To answer this question, we studied a subset of 298 cadaver kidney transplant recipients who had not suffered early graft loss and had a serum creatinine level < or = 2.0 mg/dl at 1 year after TX. The incidence of DGF (defined by dialysis during the first week after TX) in this subset was 19%. DGF was associated with cold ischemia time >24 hr (P = 0.0003) and Rej (P = 0.06). For grafts with versus without DGF, the incidence of late acute Rej (>1 year after TX) was similar. Actuarial graft survival was similar for Rej-free recipients with versus without DGF (P = 0.9) and was worse for those with Rej and no DGF (P < 0.02). Importantly, however, in our recipients who all had a serum creatinine level < or = 2.0 mg/dl at 1 year after TX, the worst long-term outcome was noted in the subgroup with both DGF and Rej (P < 0.0001). By multivariate analysis, DGF was also only a risk factor in combination with Rej (P = 0.002, relative risk = 3.7), while a 0-antigen HLA mismatch had no impact. Patient survival decreased for recipients with both DGF and Rej by univariate (P = 0.009) and multivariate (P = 0.02, relative risk = 2.9) analyses. We conclude that DGF without Rej has no impact on long-term survival. However, our data for recipients with both DGF and Rej suggest that a chronic ongoing process leads to late graft failure. Further research is necessary to identify the exact pathophysiology of this process, which appears to be, at least in part, HLA antigen independent.