D-dimer levels appear to be associated with the extent of the thromboembolic burden in patients with pulmonary embolism (PE). We therefore hypothesized that D-dimer levels at admission would be associated with prospective risk of mortality in patients with PE. We used data from 366 patients diagnosed with PE at four hospital emergency departments. A highly sensitive D-dimer test was prospectively performed at admission. The outcome was overall mortality within three months. We divided patients into quartiles on the basis of their D-dimer levels and compared mortality rates by quartile. We estimated sensitivity, specificity, and predictive values for mortality in the first and fourth quartile. Overall mortality was 5.2%. Patients who died had higher median D-dimer levels than patients who survived (4578 versus 2946 microg/l; p = 0.005). Mortality increased with increasing D-dimer levels, rising from 1.1% in the first quartile (<1500 microg/l) to 9.1% in the fourth quartile (>5500 microg/l) (P = 0.049). Sensitivity, specificity, and positive and negative predictive values of D-dimer levels <1500 microg/l to predict mortality were 95%, 26%, 7%, and 99%, respectively. Patients with PE who have D-dimer levels below 1500 microg/l have a very low mortality. Further studies must assess whether D-dimer, alone or combined with other prognostic instruments for PE, can be used to identify low-risk patients with PE who are potential candidates for outpatient treatment or an abbreviated hospital stay.