In the last twenty years, several systems have been proposed to codify renal allograft rejection. The Banff classification for kidney allograft pathology, introduced in 1993, started a new era in the standardization of criteria for rejection and for allowing uniform reporting. This consensus on allograft grading proposed a scheme to guide therapy in transplant patients and to help establish an objective rejection end point in clinical trials. This scheme, modified during 1993-1997 to address many of the criticisms, was substantially improved. Another important system of classification of allograft renal biopsies was the Cooperative Clinical Trials in Transplantation (CCTT) classification which was published in 1997. The aim of this system was to develop a schema that would be practical to implement, easy to describe to unfamiliar personnel, reproducible, with high rates of sensitivity and specificity and clinically informative (predictive of course and/or response to therapy). In March 1997, a fundamental revision of the Banff classification for acute rejection was achieved by a consensus conference for incorporating many of the strengths of the CCTT system. Some of these were the importance of vascular damage (endoarteritis, endothelial activation, fibrinoid necrosis) and interstitial hemorrhage, but not the interstitial infiltrate or tubulitis, which correlated with response to anti-rejection therapy and/or 1 year clinical outcome. The most recent modification concerns the addition of C4d-positive acute humoral rejection and the emphasis on differences between cell-mediated and antibody-mediated rejections. Future refinements of these classifications and findings of new molecular markers of allograft rejection, such as fas-ligand or granzyme-B, will help to improve diagnosis and therapy in renal transplant patients.