Background: Although glycated hemoglobin (HbA1c) was considered as a prognostic factor in some subgroup of coronary artery disease (CAD), the specific relationship between HbA1c and the long-term all-cause death remains controversial in patients with CAD. Methods: The study enrolled 37,596 CAD patients and measured HbAlc at admission in Guangdong Provincial People's Hospital. The patients were divided into 4 groups according to HbAlc level (Quartile 1: HbA1c ≤ 5.7%; Quartile 2: 5.7% < HbA1c ≤ 6.1%; Quartile 3: 6.1% < HbA1c ≤ 6.7%; Quartile 4: HbA1c > 6.7%). The study endpoint was all-cause death. The restricted cubic splines and cox proportional hazards models were used to investigate the association between baseline HbAlc levels and long-term all-cause mortality. Results: The median follow-up was 4 years. The cox proportional hazards models revealed that HbAlc is an independent risk factor in the long-term all-cause mortality. We also found an approximate U-shape association between HbA1c and the risk of mortality, including increased risk of mortality when HbA1c ≤ 5.7% and HbA1c > 6.7% [Compared with Quartile 2, Quartile 1 (HbA1c ≤ 5.7), aHR = 1.13, 95% CI:1.01-1.26, P < 0.05; Quartile 3 (6.1% < HbA1c ≤ 6.7%), aHR = 1.04, 95% CI:0.93-1.17, P =0.49; Quartile 4 (HbA1c > 6.7%), aHR = 1.32, 95% CI:1.19-1.47, P < 0.05]. Conclusions: Our study indicated a U-shape relationship between HbA1c and long-term all-cause mortality in CAD patients. Copyright © 2021 Liu, Ye, Ying, Li, Chen, Wang, Lin, Chen, Lun, Huang, Li, Xu, Tan, Chen, Liu and Liu.