Rat blood exhibited a significant quinone-dependent N-oxide reductase activity towards imipramine N-oxide. The reduction mediated by the blood proceeded in the presence of both NAD(P)H and menadione under anaerobic conditions. When menadione was replaced with 1, 4-naphthoquinone or 9, 10-phenanthrenequinone, similar results were obtained. The reduction was also mediated by the combination of rat erythrocytes and plasma. The reducing activity was inhibited by dicumarol and carbon monoxide. When boiled plasma was combined with untreated erythrocytes, the N-oxide reducing activity was abolished. In contrast, when boiled erythrocytes were combined with untreated plasma, the activity was unchanged. These results suggest that the activity is caused by the heme of hemoglobin in erythrocytes and quinone reductase in plasma. In fact, erythrocytes and hemoglobin have the ability to reduce the N-oxide when supplemented with DT-diaphorase purified from rat liver in the presence of both NAD(P)H and menadione. Hemoglobin also exhibits N-oxide reductase activity with reduced menadione (menadiol). Furthermore, hematin exhibits a significant reducing activity in the presence of menadiol. The reduction appears to proceed in two steps. The first step is enzymatic reduction of quinones to dihydroquinones by quinone reductse(s) with NADPH or NADH in plasma. The second step is nonenzymatic reduction of imipramine N-oxide to imipramine by the dihydroquinones, catalyzed by the heme group of hemoglobin in erythrocytes. Cyclobenzaprine N-oxide and brucine N-oxide are similarly transformed to the corresponding amines by the above reducing system in blood.These results suggest that blood plays an important role in the reduction of tertiary amine N-oxide to tertiary amines.