Background: Few reports attempt to validate the role of Epstein criteria in selecting patients for an active surveillance protocol. Objective: To determine the performance of the Epstein biopsy criteria for predicting pathologic end points and biochemical relapse-free survival (bRFS) in men with early stage prostate cancer (PCa) treated by radical prostatectomy (RP). Design, setting, and participants: Between October 1999 and January 2007, 746 consecutive patients were biopsied, and then underwent RP at our tertiary care institution. Two hundred sixty-eight patients met the entry criteria of Gleason 6 disease only on initial biopsy with complete pathologic information. Measurements: Primary end point was insignificant disease. Insignificant disease was defined using a classical (organ-confined, Gleason score <6, and tumor volume <0.5 cm3) and more liberal (organ-confined, Gleason <6 tumor of any volume) formulation. Secondary end points included organ-confined disease and bRFS. Results and limitations: One hundred thirty-six men (51%) met the Epstein biopsy criteria, and 167 (62%) had organ-confined cancer. Insignificant disease by the classical and liberal definitions was present in 68 (25%) and 92 (34%) patients, respectively. Cases meeting Epstein biopsy criteria were more likely to have insignificant disease by either definition (p < 0.001) and more likely to have organ-confined tumors (p < 0.001). Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) varied widely among the end points, with sensitivity (74%) and NPV (86%) best for the classical definition of insignificant disease and specificity (74%) and PPV (92%) best for organ-confined disease. The estimated 5-yr bRFS was 100% for those meeting Epstein biopsy criteria compared to 83% for those not meeting these criteria. Conclusions: The Epstein biopsy criteria predict for a high likelihood of organ-confined disease and the absence of biochemical failure up to 5 yr after RP. These criteria are insufficiently robust to predict the presence of biologically insignificant disease. © 2009 European Association of Urology.