Warfarin, like all the 4-hydroxycoumarin compounds, has an asymmetric carbon atom. The clinically available warfarin preparations consist of a racemic mixture of equal amounts of 2 distinctSandRisomers, the former being 4-times more potent as anticoagulant and more susceptible to drug interaction. Warfarin is highly water soluble and rapidly absorbed from the stomach and the upper gastrointestinal tract; its plasma concentrations peak 60 to 90 minutes after oral administration. Warfarin binds to the enzyme vitamin K 2,3-epoxide reductase in liver microsomes, stopping the cycle of vitamin K and reducing &ggr;-carboxylation of the precursors of vitamin D-dependent pro- and anticoagulant factors. A variable fraction of the binding with the target enzyme, albeit small, can be reversed by competitive displacers, such as dithiol-reducing agent activity. Differences in dithiol-reducing activity have been suggested as a contributing factor to the wide interindividual differences in sensitivity to oral anticoagulants. The anticoagulant effect is caused by a small fraction of the drug, since most (97 to 99%) is protein bound (mainly to albumin) and ineffective. Drugs that can displace the albumin binding will increase the action of warfarin, even though this effect is counteracted by a more rapid elimination of the drug. The elimination half-life of warfarin varies greatly among individuals, ranging from 35 to 45 hours; theSisomer has, however, an average half-life shorter than the R isomer.The plasma levels of vitamin K-dependent proteins are determined by a dynamic equilibrium between their synthesis and half-life times. The delay before warfarin takes effect reflects the half-life of the clotting proteins; the levels of factor VII and protein C (with shorter half-lives) are reduced earlier, reaching steady inhibited levels in about 1 day, whereas factor II takes more than 10 days.Oral anticoagulant therapy (OAT) with warfarin or other coumarin derivatives is increasingly administered to patients for primary or secondary prevention of various arterial or venous thromboembolic diseases. If in some clinical conditions OAT is given indefinitely, in others - such as venous thromboembolism or after tissue heart valve replacement - anticoagulants are usually given only for the high risk period of thrombotic complication. A recent large prospective study performed by the Italian Federation of Anticoagulation Clinics showed that about 30% of the patients who began OAT for various clinical indications stopped treatment at different times, confirming that withdrawal from OAT is an occurrence that affects a large number of patients. The expression ‘rebound phenomenon’ was adopted to indicate a hypercoagulant condition occurring after warfarin withdrawal. A possible more frequent recurrence of thromboembolism after cessation of anticoagulation became a matter of controversy and many clinical studies, mostly observational and noncontrolled, reported on the issue with inconsistent results. Most authoritative commentators agreed that rebound phenomenon, though possible, was not clinically relevant and did not differ in frequency and intensity according to mode of withdrawal. Scientific interest in the topic waned until more sensitive methods for investigating blood hypercoagulability became available.In recent years, many studies (reviewed in the text) have investigated the levels of different markers of hypercoagulability [fibrinopeptide A, activated factor VII, prothrombin fragments F1+2, thrombin-antithrombin complexes, D-dimers (DD)], consistently finding an increase in their values after cessation of anticoagulation. Changes in the levels of markers of activated blood coagulation were prospectively investigated by our group in 32 patients with venous thromboembolism who were randomly withdrawn abruptly or gradually from warfarin treatment. Our results indicate that interruption of anticoagulant treatment frequently elicits low grade activation of the haemostatic system, usually not detectable during steady-state anticoagulation. This phenomenon seems to be earlier and more intense after abrupt interruption, but also seems to occur after stepwise withdrawal. Although the activation of coagulation appears to be of limited extent and transient in most cases, it may in a few patients reach higher levels. It seems reasonable to suppose that these patients are at higher risk of thrombotic complications. We and others have suggested that gradual anticoagulant withdrawal would prevent and/or blunt the rebound phenomenon. However, the need to taper the anticoagulant dosage is still uncertain, since no convincing clinical evidence is as yet available. Specifically designed clinical trials are required to solve the issue.