PurposeTo compare the efficacies of percutaneous transhepatic obliteration (PTO) with initial afferent vein embolization (PTO-A) and with initial efferent vein embolization (PTO-E) for gastric varices (GVs). Materials and MethodsConsecutive patients undergoing PTO-A or PTO-E for type 1 isolated gastric varices or type 2 gastroesophageal varices between June 2007 and February 2023 were included. PTO-A or PTO-E was indicated when balloon-occluded retrograde transvenous obliteration failed, or there was no portosystemic shunt for the retrograde transvenous approach. PTO-A was performed for patients before March 2009 and for patients with multiple efferent channels from the GVs. PTO-E was attempted for patients after April 2009. Twenty-six patients were included, of whom 18 underwent PTO-A, and 8 underwent PTO-E. In 12 patients, PTO-E was attempted without success, and they underwent PTO-A instead. The primary outcome was successful complete eradication of GVs within 1 week after the procedure. Secondary outcomes were the GV recurrence-free period and overall survival after the procedure. ResultsTechnical success was achieved in all 26 patients, with a complete GV eradication rate of 33.3% in PTO-A and 100% in PTO-E, significantly higher in PTO-E (p = 0.002). The median recurrence-free period was 8.7 months with PTO-A and 75.1 months with PTO-E; PTO-E had a significantly longer recurrence-free period (p = 0.042). Median overall survival was 37.2 months with PTO-A and 49.9 months with PTO-E, with no significant difference (p = 0.955). ConclusionPTO-E had a better complete eradication rate and longer recurrence-free period than PTO-A for GVs.