Significant controversies persist in regard to the need for systematic biopsies in patients with serum prostate-specific antigen (PSA) levels above 4 ng/mL (Hybritech assay), especially if they show no signs of prostatic cancer on digital rectal examination (DRE) or transrectal ultrasonography (TRUS). We evaluated 565 consecutive patients referred to us for prostatism, suspicious lesions on DRE, or an elevated serum PSA level. These patients do not represent a purely screened population. A detection rate of 38.4 percent was achieved by performing directed biopsies of suspicious lesions on DRE and/or TRUS, and systematic biopsies of all patients with serum PSA levels above 4 ng/mL. Among 142 patients with serum PSA between 4.1 and 10 ng/mL, but without suspicion for cancer on DRE and TRUS (DRE- TRUS-), a large number of patients (6.2) were subjected to systematic biopsies to detect one cancer. A receiver-operating characteristics curve for PSA density (PSAD) applied to this population confirmed that the best cut-off point for biopsies was a PSAD of 0.15, below which only two of twenty-three cancers would have been missed, sparing biopsies in 77 of 142 patients. A similar approach was applied to DRE- TRUS- patients with serum PSA levels above 10 ng/mL. The number of cancers in those with serum PSA between 10.1 and 14 ng/mL was too low to establish a PSAD cut-off point. In patients with serum PSA above 14 ng/mL, the best PSAD cut-off point for biopsies was 0.3, below which two of thirteen cancers would have been missed, sparing biopsies in 19 of 39 patients. We conclude that PSAD can safely reduce the number of patients subjected to systematic biopsies without significantly compromising cancer detection.