To identify predictors of arterial graft patency, we followed up 30 internal thoracic arterial grafts and 23 right gastroepiploic arterial grafts in situ with patency documented during postoperative angiography. After 24 months of follow-up on average, repeat angiography detected that one internal thoracic artery and two gastroepiploic arteries were anatomically occluded and that the other three gastroepiploic arteries were nonfunctioning. The logistic regression model identified a relationship between graft patency and competitive flow, which was detected as stenosis in the recipient coronary arteries (coefficients, p < 0.05; model, Hosmer-Lemeshow chi2 statistic 3.59, p = 0.89). The linear regression model demonstrated that changes in graft luminal diameter correlated with competitive flow (p < 0.01), smoking history (p < 0.05), and type of arterial grafts (p < 0.001) (R2 = 0.40, adjusted R2 = 0.36). The findings suggest a temporal relationship between competitive flow and prognosis of arterial graft.