The optimal duration of anticoagulant therapy after a first episode of unprovoked pulmonary thromboembolism is not fully defined. The identification of patients more prone to recurrence would be useful in this context but is currently relatively unreliable. Perfusion lung scan (PLS) is an established approach for the follow-up of patients with pulmonary embolism to identify recurrences and to help in the diagnosis of chronic thromboembolic pulmonary hypertension. The aim of the study was to investigate the potential role of residual perfusion defects at follow-up perfusion scans in predicting pulmonary embolism recurrences. We retrospectively analyzed PLSs of 252 patients with a first episode of unprovoked, symptomatic pulmonary embolism. The agreement between two experienced readers, as assessed by the kappa test, was good, with kappa indices ranging from 0.84 (baseline scan) to 0.98 (last prerecurrence available scan). Sixteen patients developed a late (at least 1 month from the index episode) recurrence identified through the appearance of (a) new perfusion defect(s) not matching radiograph alterations. When patients were divided based on the presence or absence of at least two unperfused segments at the 6-month follow-up lung scan, the probability of recurrence was significantly higher in the latter (P = 0.03 by log-rank test). The use of persistent perfusion defects at follow-up PLS as a guide to determine optimal duration of anticoagulant therapy after a first, unprovoked episode of acute pulmonary thromboembolism is a viable strategy that should be further investigated.