Venous thromboembolism is the most common reason for readmission after total knee arthroplasty. Prospective contrast venography was conducted from 1984 to 2003 in 1321 patients undergoing total knee arthroplasty. Patients with deep venous thrombosis or pulmonary embolism were treated with warfarin; those with negative venograms received no further anticoagulation. From 1984 to 1992, patients not completing venography were discharged without warfarin; since 1993 patients without venography received warfarin for 6 weeks. Readmission for deep venous thrombosis, pulmonary embolism, or bleeding was tracked for 6 months. Venography was completed in 810 patients; 343 (42.3%) had deep venous thrombosis. Readmission for venous thromboembolism occurred in 0.6% of patients after total knee compared with 1.62% after total hip arthroplasty. Following total knee arthroplasty, patients discharged on warfarin (target INR 2.0) had a 0.21% readmission rate compared with 1.05% for patients with negative venograms discharged without further anticoagulation. One patient suffered a fatal pulmonary embolism after negative venography and no outpatient prophylaxis. Secondary prophylaxis with extended warfarin therapy reduced venous thromboembolism-related readmission. Surveillance venograms were a poor predictor of ultimate thromboembolism risk and need for extended anticoagulation therapy. We therefore recommend extended warfarin prophylaxis for all patients after hospital discharge following total knee arthroplasty.