Excluding or confirming pulmonary embolism remains a diagnostic challenge. In elderly patients pulmonary embolism is associated with substantial co-morbidity and mortality, and many elderly patients with suspected pulmonary embolism are inpatients. The safety and efficacy of the combination of a clinical probability (CDR) and d-dimer test in excluding pulmonary embolism in this group is unclear. We retrospectively analysed data of two prospective studies of consecutive in-and outpatients with suspected pulmonary embolism. The patients were categorized into three age groups: <65 years, 65-75 years and >75 years. The sensitivity, negative predictive value and the proportion of patients with the combination of a non-high CDR score according to Wells (< or = 4) and a normal d-dimer result were calculated for each group. In 747 consecutive patients with suspected pulmonary embolism, sensitivity and negative predictive value of a non-high CDR and a normal d-dimer result in outpatients (n=538) of all age categories (<65 years, 65-75 years and >75 years) were both 100%. These tests were, however, less reliable for inpatients(n=209), irrespective of their age (sensitivity 91% [ CI: 79-98%], negative predictive value 88 % [CI: 74-96%]. The proportion of both in-and outpatients >75 years with a non-high CDR and a normal d-dimer concentration was only 14%, whereas this was 22% in patients 65-75 years and 41% among in-and outpatients <65 years, respectively. In elderly outpatients the combination of a non-high CDR and a normal d-dimer result is a safe strategy to rule out pulmonary embolism. However, in inpatients this algorithm is not reliable to safely exclude pulmonary embolism. In addition, the proportion of patients >65 years in which this strategy excludes pulmonary embolism is markedly lower compared to younger patients.