To evaluate the diagnostic power of penile/brachial index (PBI) in patients studied for male sexual impotence, we prospectively interviewed 503 patients referred to our vascular laboratory for PBI measurements. Since the predictive values of diagnostic tests are affected by the prevalence of the disease in the population studied, we calculated the independent likelihood ratio for various PBI levels obtained. For the purpose of this analysis, "organic" impotence was defined as occurring in patients who had no nocturnal erections, whereas "clinical" impotence was defined as occurring in those patients who could not achieve penetration during intercourse. The patients were divided into four groups by the presence or absence of risk factors, including peripheral vascular disease (PVD) and diabetes mellitus. Patients taking medications that may affect potency (n = 175) were excluded from this analysis. The results show that the predictive power of PBI is less in diabetic patients with PVD and least in those without either PVD, diabetes, or drugs. In patients with PVD but no other risk factors, PBI is highly diagnostic with a sharp cut-off point at 0.6. Thus, the diagnostic power of PBI can be improved by considering the risk factors in the patients studied.