Aims Impaired fasting glucose (IFG) below the diagnostic threshold for diabetes mellitus (DM) is associated with macrovascular pathology and increased mortality after percutaneous coronary interventions. The study goal was to determine whether pre-operative fasting blood glucose (fB-glu) is associated with an increased mortality after coronary artery bypass grafting (CABG). Methods and results During 2001–03, 1895 patients underwent primary CABG [clinical DM (CDM) in 440/1895; complete data on fB-glu for n =1375/1455]. Using pre-operative fB-glu, non-diabetics were categorized as having normal fB-glu (<5.6 mmol/L), IFG (5.6≤fB-glu<6.1 mmol/L), or suspected DM (SDM) (≥6.1 mmol/L). fB-glu was normal in 59%. The relative risks of 30 day and 1 year mortality compared with patients with normal fB-glu was 1.7 [95% confidence interval (CI): 0.5–5.5] and 2.9 (CI: 0.8–11.2) with IFG, 2.8 (CI: 1.1–7.2) and 1.9 (CI: 0.5–6.3) with SDM vs. 1.8 (CI: 0.8–4.0) and 1.6 (CI: 0.6–4.3) if CDM, respectively. The receiver operator characteristic area for the continuous variable fB-glu and 1 year mortality was 0.65 ( P =0.002). Conclusion The elevated risk of death after CABG surgery known previously to be associated with CDM seems also to be shared by a group of similar size that includes patients with IFG and undiagnosed DM.