In renal transplantation, the long-term graft survival rate has not been improved. Until now, the differences between late graft loss and long-term graft survival have still not been estimated thoroughly. We have attempted to define clinical risk factors and parameters for late graft loss by comparing the differences in these two groups. Data from the Osaka University Database were assessed on 156 renal allografts during a 7-yr period. Thirty-six patients comprised the late graft loss group (patients in this group had graft function without need for dialysis for more than 3 yr post-transplantation, afterwards lost the allograft: 'loss group'). One hundred and twenty patients comprised the long-term graft survival group (patients in this group had graft function without need for dialysis until 31 December 1999: 'survival group'). Various immunological and non-immunological parameters were included in an univariate regression analysis. This analysis showed that donor age (P < 0.01), HLA mismatch number (P < 0.01) and a repeat of acute rejection (P < 0.01) were significant factors. Serum creatinine levels at 3 months (P = 0.01), proteinuria at 1 yr (P < 0.01) and antihypertensive treatment at 2 yr (P = 0.03) after transplantation were predictive of the risk of late graft loss. CsA trough concentration at 3-6 months (P < 0.05) and body mass index increase at 1 yr (P = 0.046) were elevated in the loss group. These results from a single centre suggest that immunological as well as non-immunological factors are associated with the pathogenesis of late graft loss.