There is accumulating evidence regarding the overuse of computed tomography pulmonary angiography (CTPA) to exclude pulmonary embolism (PE). We evaluated the yield of CTPA studies performed at our tertiary care hospital between April 2008 and March 2010 for emergency patients (ED), inpatients (INPT), and intensive care unit inpatients (ICU). For each patient group, we also compared CTPA positivity rates among the following: daytime and on-call studies, 1 year before and after institution of an Emergency Radiology division, interpreting thoracic and non-thoracic radiologists, and individual emergency physicians. Patients with a history of PE and indeterminate studies were excluded. The diagnosis of PE was based on the radiology report. D-dimer values obtained within 24 h prior to CTPA were recorded. A total of 3,571/4,757 CTPA studies satisfied the inclusion criteria. The fraction of positive studies was 252/1,677 (15.0 %) ED, 255/1,548 (16.5 %) INPT, and 62/346 (17.9 %) ICU. There was no difference in yield between patient groups, daytime vs. on-call studies, before vs. after instituting an emergency radiology division, and thoracic vs. non-thoracic radiologists (p > 0.05). For individual emergency physicians, the mean CTPA positivity rate was 15.4 % but varied considerably (σ = 8.5 %, range, 0-38.5 %). In comparison to other recent studies, our yield of ED CTPA is relatively high but varied widely among individual emergency physicians. While the reasons for such differences require further investigation, our results reinforce the importance of a strong clinical assessment in the work-up of suspected PE.