Prostate cancer is the fifth most frequent cancer in the world. However, none of the actual prognostic factors provide a valid index for predicting patient outcome. Here, we evaluate the two-dimensional vascularity in primary prostate tumors and surrounding non-tumoral parenchyma by means of fractal geometry, and assess any correlations between the results and some clinical and pathological parameters of prostate carcinoma. Prostate sections from 27 carcinoma patients were treated with CD34 antibodies. Two >10mm(2) areas of tumoral and surrounding non-tumoral parenchyma were digitized using an image analysis system that automatically quantified the fractal dimension of the vascular surface. Data were correlated with patient's age, PSA level, clinical and pathological stage, Gleason score, tumor volume, vascular invasion, surgical margins, and biochemical relapse. Two groups of patients were distinguished on the basis of whether the fractal dimension of their tumoral vascular surface was higher (group 1) or lower (group 2) than that of the surrounding non-tumoral parenchyma. Statistically significant between-group differences were found in terms of serum PSA levels (p=0.0061), tumor volume (p=0.0017), and biochemical relapse (p=0.031). The patients in group 2 had a poorer outcome. Our findings suggest a group of prostate cancer patients with a poor outcome, and the vascular surface fractal dimension as a helpful geometrical index in clinical practice.