In this study, I have estimated the percentage of tumor by visual estimate in 447 prostatectomy specimens from American Veterans Affairs patients and related this measurement to overall survival. Although percentage of tumor was significantly related to the serum prostate-specific antigen level, tumor stage, and Gleason score-that is, it was not statistically independent from these-it was more closely associated with overall survival than any of them. Altogether, 2 variables available at the time of the prostatectomy related to survival: patient age (P = .0032; Cox proportional hazards model analysis) and percentage of tumor (P = .0013; Cox model). Patient age undoubtedly reflects the combination of comorbidities and general expected length of life. Percentage of tumor, by contrast, seems to efficiently reflect any undue hazard for early death due to prostate cancer. The results suggest that percentage of tumor is a useful prognostic variable for understanding risk of early death after prostatectomy.