The purpose of this study was to prospectively compare the efficacy and controllability of pushable coil and detachable coil during embolization of gastroduodenal artery (GDA) while performing percutaneous implantation of port-catheter system for hepatic artery infusion chemotherapy. Fifty patients (M:F = 42:8, age: 31-81 years) with advanced hepatocellular carcinoma undergoing port-catheter system implantation were randomized into pushable coil group and detachable coil group. During catheter fixation, GDA was embolized as close to the origin as possible. Success rate, number of coils used, number of coils removed due to malposition after deployment, time to occlusion, uncoiled GDA length, pushability, and complications were compared. Pushability was graded as no tension, slight tension, and difficult to advance. Embolization was successful in 49 patients. One failure resulted from repeated regurgitation of pushable coil into hepatic artery. Number of coils used and removed coils, time to occlusion, and uncoiled GDA length were 1-3 (mean 2.32), 5 coils in 3 patients, 4-20 min (mean 8.00), and 0-15.0 mm (mean 3.36) in pushable coil group, and 1-5 (mean 2.12), 2 coils in 2 patients, 2-15 min (mean 7.40), and 0-10.2 mm (mean 2.92) in detachable coil group, respectively, without significant difference. Pushability was no tension (n = 24) and slight tension (n = 1) in pushable coil group and no tension (n = 16), slight tension (n = 7), and difficult to advance (n = 2) in detachable coil group. One hepatic artery dissection occurred in the failed case during coil removal. Pushable coils and detachable coils had similar efficacy and controllability during GDA embolization, although there was a trend favoring detachable coil.