To evaluate the impact of tumor length and fraction of positive biopsy cores on overall survival, I used the data for 526 patients with prostate cancer. Median follow-up in patients not observed until death was more than 6 years. In a Cox model analysis that included age, serum prostate-specific antigen (PSA) level, grade, and fraction of positive cores, tumor length was the most closely associated with overall survival time (P=6 x 10(-5)); however, the impact of tumor length was mostly for a subset of men with tumors measuring more than 20 mm. Patient age, serum PSA level, Gleason score, fraction of cores with tumor, and tumor length were all significantly codependent variables. For routine cases of prostate cancer, measuring tumor length in the needle cores may be unnecessary. Tumor length may assist studies of long-term outcomes or treatment trials in prostate cancer by reducing baseline variance better than other prognostic variables. For the few patients with unusually large amounts of tumor in biopsy specimens, tumor length may provide a concise indicator for the likelihood of an adverse outcome, especially when the values of other prognostic variables appear by themselves to be less ominous.