The purpose of this study is to retrospectively examine the patterns of diagnostic modality utilization in the setting of suspected acute pulmonary embolism (PE) by physicians in the emergency department (ED) of a major academic tertiary care center. All patients (n=180) with a hospital discharge diagnosis of PE that were admitted through the ED during 1997 to 1999 were included. Overall test frequency and distribution, ancillary testing used when the initial ventilation-perfusion (V/Q) scan or spiral computed tomography (sCT) angiogram was nondiagnostic, concordance of V/Q and sCT, and sensitivities of D-dimer assays, arterial blood gas analysis, and deep venous imaging were assessed. There was a wide discrepancy in practice, most pronounced when first-line radiological studies were nondiagnostic. There was a poor correlation of PE detection between sCT and V/Q. D-dimer assay, arterial blood gas analysis, and deep venous imaging are insensitive tests for PE, and "gold standard" pulmonary angiography is rarely used.