A limited ultrasound (US) confined to the popliteal and femoral veins is usually performed to detect deep vein thrombosis (DVT) in patients with clinically suspected acute pulmonary embolism (PE). Our objective was to assess the diagnostic accuracy of complete lower limb US examining both the proximal and distal veins in this setting. In this prospective study, 210 consecutive patients were included. Complete US was performed by independent operators and compared blindly with a reference strategy combining clinical probability, ventilation perfusion scan and pulmonary angiography to a three-month clinical follow-up. Simultaneously, VIDAS D-dimer (DD) assay and helical computed tomography (HCT) of the lungs were assessed independently and blindly. PE was present in 74 patients (35%). Complete US detected DVT in 91 patients (43%), proximal in 51 and distal in 40. Sensitivity and specificity with a 0.95 confidence interval were respectively 0.93 [0.85 - 0.97] and 0.84 [0.77 - 0.89]. Limited US detected DVT in only 46 patients (22%). Sensitivity and specificity were respectively 0.55 [0.44 - 0.66] and 0.96 [0.92 - 0.98]. For DD they were 0.92 [0.83 - 0.96] and 0.24 [0.17 - 0.32] and for HCT 0.84 [0.73 - 0.90] and 0.87 [0.80 - 0.92]. Complete lower limb US has higher sensitivity and capacity to exclude PE than limited US, but a slightly lower specificity. Complete US results also compared favourably with those of HCT and DD. The utility of including this method in diagnostic strategies for PE needs to be assessed in cost-effectiveness analysis and in outcome studies.