The role of the liver biopsy for establishing a diagnosis of liver allograft rejection was investigated by examining under code 151 liver biopsies from 32 recipients and correlating the results with the clinical diagnosis determined by chart review. One-hundred-and-four biopsies were obtained to evaluate the cause of liver dysfunction and 47 were performed on a weekly protocol basis independent of the clinical status. A diagnosis of "definite histologic rejection" was made if a biopsy contained portal triaditis, bile duct injury, and endothelialitis, while biopsies were said to be suspicious for rejection if portal triaditis and bile duct injury were evident. A diagnosis of rejection using clinical and biochemical findings and response to therapy was made in association with 37 biopsies, of which 16 had definite histologic rejection and 13 were suspicious for rejection. Of the 114 biopsies performed in patients without clinical rejection, 71 had histology suspicious for rejection. Thus, biopsy was sensitive for the detection of rejection (78%), but was not very specific (33%). The positive predictive value was low (28%), while the negative predictive value was high (83%). These results indicate that liver biopsy is useful to evaluate graft dysfunction, to exclude rejection, to confirm clinically suspected rejection, and to diagnose other problems that can affect allograft function. Protocol biopsies appear to have little value.