OBJECTIVE: We investigated the association between adherence to Mediterranean diet and fasting indices of glucose homoeostasis, in a Greek adult population. METHODS: During 2001-2002 we randomly enrolled 1514 men and 1528 women (18-89 years old) without history of CVD, from the Attica area. Diabetes mellitus (type 2) and impaired fasting glucose (IFG) were defined according to the established ADA criteria. Insulin resistance was evaluated by HOMA-IR. Dietary habits were assessed through a validated food frequency questionnaire and a diet score (range 0-55) was developed (higher values means greater adherence to the Mediterranean diet). RESULTS: The overall prevalence of diabetes type 2 was 7.9% in men and 6.0% in women (P = 0.05). Mean diet score was 26.3 +/- 6.8 in normoglycemic, 25.7 +/- 6.4 in IFG and 22.2 +/- 5.8 in diabetic subjects (p < 0.001). In normoglycemic subjects who were in the upper tertile of the diet score we observed 7% lower glucose (p < 0.05), 5% lower insulin (p < 0.05) and 15% lower HOMA-IR (p < 0.01) levels compared to subjects in the lower tertile of the diet score. Additionally, in diabetic/IFG participants who where in the upper tertile of the diet score we observed 15% lower glucose (p < 0.05), 15% lower insulin (p < 0.05) and 27% lower HOMA-IR (p < 0.01) levels compared to those in the lower tertile. However, multiple regression analysis, adjusted for age, sex, BMI, waist-to-hip ratio, physical activity, smoking status, and presence of hypertension and hypercholesterolemia, confirmed the previous associations in normoglycemic, but not in diabetic/IFG people. CONCLUSION: An inverse association was observed between adherence to Mediterranean diet and indices of glucose homeostasis, only in normoglycemic people.