Objectives. To retrospectively study evaluated prostate-specific antigen (PSA) and four PSA indices as criteria for performing a repeat biopsy when the initial biopsy findings are negative for cancer. Methods. One, two, or more repeat biopsy sessions were performed on 193, 54, and 14 men, respectively, all of whom had an initially negative biopsy at our institution. We compared the usefulness of PSA, PSA density, age-referenced PSA, volume-referenced PSA, and PSA velocity for predicting the presence of cancer. Results. Overall, 51 men (26%) were found to have cancer on repeat biopsy. Cancer was found in 17% (33 of 193) of the men on the first repeat biopsy and 26% (14 of 54) of the men who had a second repeat biopsy. Of all the indices, PSA and volume-referenced PSA had the highest sensitivity, missing the fewest cancers; however, this was achieved at the expense of saving the least number of biopsies. We evaluated the usefulness of indices among 9 patients who had a normal PSA at the initial biopsy; volume-referenced PSA was the earliest predictive index indicating that a repeat biopsy be performed. Analysis of the area under the receiver-operating characteristic curves revealed no significant advantage for any index over PSA alone in determining who should undergo a repeat biopsy. Conclusions. In determining the need for repeat biopsies, the established PSA threshold value of 4 ng/mL is equivalent or superior to age-referenced PSA, volume-referenced PSA, PSA density, and PSA velocity. However, volume-referenced PSA has the potential to be the earliest index predicting cancer in men with normal PSA level. (C) 1997, Elsevier Science Inc. All rights reserved.