The objective of the study was to assess the clinical validity of a non-invasive diagnostic strategy for acute pulmonary embolism using clinical assessment combined with both ELISA D-dimer and complete lower limb ultrasound (US) examination of proximal and distal veins, before single-detector helical computed tomography (CT) of pulmonary arteries. We expected the strategy to have a high diagnostic exclusion power and to safely decrease the number of CT scans. This prospective, multicenter outcome study included 274 consecutive outpatients. All underwent a priori clinical probability, D-dimer and bilateral complete lower limb US assessments. Only patients with a high clinical probability and both tests negative, or positive D-dimer and negative US assessments, underwent CT. This was deemed necessary in 114 patients (42%). At baseline, venous thromboembolism (VTE) was detected in 110 patients (40%), either by US showing proximal (n=65) or distal (n=36) thrombosis, or by CT (n=9). Anticoagulant was withheld in the remaining patients with negative results in both D-dimer and US but a non-high clinical probability (n=59), or in both US and CT (n=90), or with negative US (n=6) and inadequate CT (n=9). All patients underwent a three-month clinical follow-up. VTE occurred in one patient with inadequate CT, yielding an incidence of 0.6% [95% confidence interval: 0.1-3.4]. No patient died from VTE or had major bleeding. Using clinical probability, ELISA D-dimer and complete US before helical CT is a safe strategy resulting in a substantial reduction in CT scans.