Potential reasons for discordance between the Gleason score in biopsies and surgical specimens are: 1) pathological interpretation bias, and 2) sampling effects. The importance of sampling effects in grading errors was examined in a series where the number of biopsy cores obtained was high. Biopsies were obtained using a technique whereby 18 directed cores were systematically obtained and mapped out within the gland. Gleason scores from biopsies and matched prostatectomy specimens were compared among 28 consecutive patients with localized prostate cancer. A pooled database from 10 series (n = 2,687) served as a baseline for comparison in the accuracy of Gleason score grading. With the present biopsy technique, an exact Gleason score match was achieved in 57% of cases, compared with the pooled database (PD) mean of 42% (P = 0.055), and was within 1 point in 93% of cases compared with 78% (PD) (P = 0.029). Upgrading of biopsies was seen in 35% of cases, compared with 43% (PD) (P = 0.19). With respect to Gleason score 7, an exact match was present in 78% of cases, compared with 63% (PD) (P = 0.17), and upgrading was 0%, compared with 20% (PD) (P = 0.07). The data suggest a significant reduction in grade errors by minimizing sampling effects, one that it is of the same order of magnitude as the reduction achieved from consensus pathologic evaluation. In our study, seven patients (25%) would have had their cancers missed altogether with sextant biopsies. Sampling effects may contribute significantly to grading errors in prostate needle biopsies, although a larger study is needed to confirm this. A methodology which adopts a higher number of cores combined with a consensus pathologic evaluation could potentially reduce grading errors substantially. The optimal number of cores remains to be determined in a larger study. Int. J. Cancer (Radiat. Oncol. Invest.) 90, 326-330 (2000).