The effectiveness of new diagnostic tools for suspected pulmonary embolism (PE), such as clinical probability assessment, plasma D-dimer (DD) measurement and lower limb venous compression ultrasonography (US), has not been specifically studied in patients with a suspected PE occurring during hospital stay. This study applied a sequential, decision analysis-based strategy adding these instruments to a ventilation/perfusion lung scan in a cohort of 114 consecutive inpatients clinically suspected of PE in order to establish in how many patients a pulmonary angiogram could thereby be avoided. A definitive diagnosis could be established by the noninvasive protocol in 61% of these patients: normal/near-normal lung scan, 14%; high probability lung scan, 19%; clinical probability combined with lung scan result, 18%; and US, 8%. Specificity of DD was only 7% and contributed to the exclusion of PE in only two patients. Pulmonary angiography was required in 39% of patients. The 3-month thromboembolic risk in patients in whom PE was excluded by the diagnostic process was 0% (95% confidence interval 0-4.9%). In conclusion, a noninvasive work-up for suspected pulmonary embolism is effective in hospitalized patients, allowing to forego angiography in 61% of them, and it appears to be safe, although this should be further investigated. In contrast to outpatients, D-dimer measurement appears to be useless in hospitalized patients.