The concept of the 'French paradox' has been recently challenged. As it is difficult in a short period to produce direct clinical evidence of the protective effect of red wine on thrombosis, we evaluated such a possibility in an experimental model mimicking the conditions of the 'French paradox'. Normolipidemic rats (FNL) were fed a standard diet or a 2% cholesterol-rich-diet (Ch-rich-diet) for 5 months: the latter was given either alone (FNL + D) or in combination with 'alcohol-free' red wine (FNL + D + 5 W). Arterial thrombosis was measured as the occlusion time (OT) of an artificial prosthesis inserted into the abdominal aorta. Lipid levels, platelet adhesion to fibrillar collagen, factor VII (FVII) clotting activity and fibrinogen levels were also measured. Compared to animals fed a standard diet, Ch-rich diet induced in FNL rats a several-fold increase in lipids and FVII levels with a concomitant significant increase in both thrombotic tendency (shortening of the OT) and platelet adhesion. 'Alcohol-free' red wine supplementation almost completely reverted the prothrombotic effect of the Ch-rich-diet. Indeed, the OT was prolonged from 78 +/- 3 to 122 +/- 10 h (P < 0.01), while platelet adhesion to fibrillar collagen was reduced from 49 +/- 3.5% to 30 +/- 2.8%. Neither the increase in lipid levels induced by Ch-rich diet nor FVII or fibrinogen levels were modified by wine supplementation. In conclusion, in experimental animals, this study supports the concept of the 'French paradox' that regular consumption of wine (rather than alcohol) was able to prevent arterial thrombosis associated with dietary-induced hypercholesterolemia, an effect mediated by downregulation of platelet function.