An article in the August 12/26, 2002, issue of the ARCHIVES compared warfarin with daily subcutaneous low-molecular-weight heparin (LMWH) in the secondary prevention of venous thromboembolism (VTE) in patients with cancer.1 The authors concluded that the use of warfarin was far more dangerous because of bleeding complications. I submit that it is more likely that the excess bleeding in the warfarin group was caused by allowing the patients' international normalized ratio (INR) to be rated as acceptable up to 3.0. Although this level of anticoagulation has been approved in the general medical community, it is my opinion from 25 years of using warfarin that an INR of 3.0 is far too dangerous. The study in question clearly demonstrates this point in that the INR was only in the "therapeutic" range on 41% of determinations. In addition, most of the serious bleeding occurred when the INR had slipped above 3.0. The level of anticoagulation in this study was controlled by principal investigators or primary care physicians, but these doctors were unfortunately following the current guidelines indicating that an INR of 3.0 requires no reduction in warfarin dosage. I believe this is a very serious mistake, and one that killed 8.5% of their patients in only 3 months. Unfortunately, anticoagulation clinics using these guidelines would do no better, and in fact might even have worse results. The technicians in these clinics have no direct knowledge of the patients in terms of their blood pressure history, prior cerebrovascular accidents, or, for example, an old bleeding ulcer from 20 years prior. Our local medical school runs such a clinic, and I shudder every time one of my patients is directed there for anticoagulation control. I suggest that in the real world physicians aim for an INR of 2.0, realizing that many times the actual value will be far higher, though at times somewhat lower. Lastly, in view of recent questions relating to possible bias in the selection of patients in medical research, I would like to ask who paid for this study?