D-dimer can be used to exclude acute pulmonary embolism (PE) for its high negative predictive value (NPV). Also, it is a predictor of recurrent venous thromboembolism (VTE) after anticoagulation withdrawal. The aim of the present study was to assess the predictive value of D-dimer for recurrent VTE when tested at hospital discharge. Plasma D-dimer levels were repeatedly measured at hospital discharge in 204 consecutive patients with the first episode of acute pulmonary embolism. Patients were categorized to two groups by D-dimer levels at hospital discharge and followed up at 3, 6, and 12 months and yearly thereafter. The primary end point was symptomatic, recurrent fatal or nonfatal VTE. D-dimer levels were persistently abnormal in 66 patients (32%). After 31±19 months follow-up, patients with persistently abnormal D-dimer level levels showed a higher rate of of recurrent VTE (14 patients, 21%) compared to those with D-dimer regression (8 patients, 6%) (P = 0.001). At the multivariate analysis, after adjustment for other relevant factors, persistently abnormal D-dimer level levels were an independent predictor of recurrent VTE in all subjects investigated, (hazard ratio, 4.10; 95% CI, 1.61-10.39; P = 0.003), especially in those with unprovoked PE (hazard ratio, 4.61; 95% CI, 1.85-11.49; P = 0.001). The negative predictive value of D-dimer was 94.2 and 92.9% in all subjects or those with unprovoked PE, respectively. Persistently abnormal D-dimer level levels at hospital discharge have a high negative predictive value for recurrence in patients with acute pulmonary embolism, especially in subjects with an unprovoked previous event.