Real-time ultrasonography (US) was performed on the allografts of 100 consecutive renal transplant recipients at the time of allograft biopsy. Evaluation of the sonograms included the grading of parameters previously demonstrated to be indicative of allograft rejection. The appearance of the renal sinus fat, allograft size, corticomedullary ratio, sharpness of the corticomedullary junction, medullary conspicuity, presence of focal parenchymal abnormalities, and thickening of the pelvic or infundibular wall were individually evaluated. The authors correlated the US and the histopathologic findings. While the accuracy of a positive prediction of rejection was relatively high (83%-90%), this result is influenced by the relatively high prevalence of rejection in the biopsy group (83%). Accuracy of a negative prediction was uniformly low (17%-30%). Mild rejection was difficult to differentiate ultrasonographically from no rejection, although severe rejection could usually be differentiated from mild or no rejection, particularly in patients with the interstitial type of rejection.