Multiple logistic regression models were used in a cross-sectional study to determine the relation of fasting glycemia to angiographically assessed coronary artery disease (CAD) in 569 men (aged 18 to 69 years) who were stratified according to fasting blood glucose concentrations (<6.1 mmol/L, and 6.1 to 6.9 mmol/L or 110 to 124 mg/dl), waist circumference (<90 vs >or=90 cm), and fasting triglyceridemia (<2.0 vs >or=2.0 mmol/L or <177 vs >or=177 mg/dl). For this purpose, nondiabetic impaired fasting glucose was defined as from 6.1 to 6.9 mmol/L (110 to 124 mg/dl) compared with 250 normoglycemic controls (fasting glycemia <6.1 mmol/L or <124 mg/dl) without history of CAD. In the absence of "hypertriglyceridemic waist," impaired fasting glucose was not predictive of CAD. However, the risk of CAD was markedly higher among subjects characterized by both the hypertriglyceridemic waist phenotype and the presence of impaired fasting glucose (odds ratio 8.5, 95% confidence intervals 3.5 to 20.4; p <0.05) compared with the normoglycemic group with low waist circumferences and triglyceride levels. Thus, the results of the present study emphasizes the importance of other underlying metabolic abnormalities, such as abdominal obesity and related atherogenic dyslipidemia, in the modulation of the CAD risk associated with hyperglycemia.