BACKGROUND: Previous studies are mixed as to whether patients with unprovoked pulmonary embolism (PE) have a higher rate of venous thromboembolism (VTE) recurrence after anticoagulation is discontinued than patients with unprovoked deep vein thrombosis (DVT). OBJECTIVES: To determine whether patients with unprovoked PE have a higher rate of VTE recurrence than patients with unprovoked DVT in a prospective multicenter cohort study. PATIENTS/METHODS: Six hundred and forty-six patients with a first episode of symptomatic unprovoked VTE were treated with heparin and subsequent oral anticoagulation for 5-7 months, and were followed every 6 months for recurrent VTE after their anticoagulant therapy was discontinued. RESULTS: Of 646 patients, 194 had isolated PE, 339 had isolated DVT, and 113 had both DVT and PE. After a mean of 18 months of follow-up, there were 91 recurrent VTE events (9.5% annualized risk of recurrent VTE in the total population). The crude recurrent VTE rate for the isolated PE, isolated DVT and DVT and PE groups were 7.7%, 16.5% and 17.7%, respectively. The relative risk of recurrent VTE for isolated DVT vs. isolated PE was 2.1 (95% confidence interval 1.2-3.7). CONCLUSIONS: This study has demonstrated that patients with a first episode of unprovoked isolated DVT are 2.1 times more likely to have a recurrent VTE episode than patients with a first episode of unprovoked isolated PE. These findings need to be considered when determining the optimal duration of anticoagulant therapy for patients with unprovoked VTE.