BACKGROUND We reviewed 843 first cadaver kidney transplants carried out consecutively at our center to examine the effect on long-term graft survival of the duration of delayed graft function (DGF), defined as the time taken for the kidney to attain the threshold of a Cockcroft calculated creatinine clearance (cCCr) > or = 10 ml/min. METHODS Using a multivariate Cox survival analysis we evaluated the consequences of DGF on allograft survival, and then by regression analysis identified the factors contributing to the occurrence of DGF. Finally, using a Kaplan Meier analysis we compared the profiles of graft failure according to the duration of DGF. RESULTS Defining DGF in terms of cCCr rather than necessity for dialysis after transplantation allowed better prediction of long-term graft loss. Indeed, patients with a Cockcroft-based DGF > six days who did not require dialysis (12%) had a significantly poorer long-term graft outcome than those with a DGF < or = six days. Furthermore, we showed that a DGF of six days could be taken as a cut-off point that marked a significant difference in the long-term graft survival rate (P < 0.0001). Surprisingly, further extension of the duration of DGF > six days was not associated with further worsening of graft survival (except in DGF > 30 days). CONCLUSIONS Our results suggest a threshold effect in the lesions that ultimately results in long-term functional deficiency. In addition, we show that the need for dialysis is not an adequate criterium for DGF in terms of long-term outcome prediction.