We have previously shown that long-term treatment with eicosapentaenoic acid (EPA) improves endothelium-dependent vasodilation of the atherosclerotic arteries in both animals and humans. The aim of the present study was to examine whether EPA treatment also improves metabolic vasodilation evoked by exercise in patients with coronary artery disease (CAD). Forearm blood flow (FBF) was measured by strain gauge plethysmography in 10 patients with stable CAD, before and 3 months after oral treatment with EPA (1,800 mg/kg). FBF was measured at rest and during intra-arterial infusion of acetylcholine or sodium nitroprusside, before and after intra-arterial infusion of NG-monomethyl-L-arginine (L-NMMA, an inhibitor of nitric oxide (NO) synthesis). A rhythmic handgrip exercise was also performed for 3 min before and after L-NMMA, and FBF was measured for 3 min just after the handgrip exercise. These protocols were repeated after the long-term treatment with EPA for 3 months. The long-term treatment with EPA significantly improved the FBF responses to acetylcholine (p < 0.01), which was significantly reduced by acute administration of L-NMMA (p < 0.01). By contrast, the EPA treatment did not affect the endothelium-independent responses to sodium nitroprusside. Metabolic increases in FBF caused by the handgrip exercise were not significantly decreased by L-NMMA before the EPA treatment. The EPA treatment significantly augmented the exercise-induced increases in FBF (p < 0.05) and L-NMMA acutely abolished this augmentation (p < 0.01). These results indicate that long-term treatment with EPA improves both endothelium-dependent and exercise-induced forearm vasodilations in patients with CAD and that NO is substantially involved in the EPA-induced improvement of the FBF responses in patients with CAD.