Despite careful monitoring of oral anticoagulant treatment (OAT), some international normalized ratio (INR) for prothrombin time values will fall outside the therapeutic range. Considerable changes in serial INR results from OAT patients may be caused by random fluctuation alone, and, for statistical reasons, a fraction of the INR values will fall outside therapeutic range and interfere with dose adjustments. On the basis of therapeutic intervals and statistical evaluation of reference changes, we suggest and discuss an alternative method for interpretation of serial INR measurements. Retrospective evaluation of serial measurements of INR from OAT patients revealed an "overshooting" phenomenon. When a dose was adjusted on the basis of insignificant change in INR value, the subsequent INR value generally fell in the opposite direction. If a further change of dose was initiated because of the new INR value, a similar course in the opposite direction was observed. This "ping-pong" effect renders patients in a fluctuating state of anticoagulation and may introduce increased risk of complications. The suggested method provides an objective criterion for dose adjustments in OAT, which should reduce patients' risk.