BACKGROUND AND PURPOSE: Gonadal vein angioembolization is a successful means of primary and salvage treatment for symptomatic varicoceles. We aim to investigate angiographic findings during embolization of primary varicoceles vs those with failed surgical ligation. PATIENTS AND METHODS: Between 1992 and 2010, 106 cases referred to our interventional radiologists for primary or salvage varicocelectomy were reviewed. These patients underwent venography and gonadal vein embolization using a combination of embolization coils and vascular plugs. All images were reviewed by an interventional radiologist to determine the anatomic etiology of the varicocele. Primary and salvage embolization cohorts were compared using t test and chi-square analyses for continuous and categorical variables, respectively. Angiographic parameters were analyzed using univariate and multivariable regression models to determine significance in predicting primary vs salvage status. RESULTS: Of the 106 patients, 46 patients (57 testicles) underwent primary and 60 patients (62 testicles) underwent salvage embolization. The salvage cohort of patients was younger (P<0.001) and comprised more solely left-sided pathology (P=0.002). An equivalent number of gonadal vein divisions and proportion of patent gonadal veins was found. However, there was a significantly higher proportion (27.8% vs 6.7%) of inguinal collateral vessels and combined presence of inguinal and retroperitoneal collateral veins (8.5% vs 2.1%) identified in the cohort undergoing embolization after failed surgical varicocelectomy. Presence of inguinal collaterals (P=0.008) as well as presence of both inguinal and retroperitoneal collaterals (P=0.038) on multivariable regression analysis revealed both as independent prognosticators of salvage status. CONCLUSION: Recurrence after surgical varicocelectomy is associated with increased inguinal collaterals. The pitfall presented by this anatomic variant to surgical ligation may be successfully managed with selective gonadal vein embolization.