Variations in emergency department (ED) syncope management have not been well studied. The goals of this study were to assess variations in management, and emergency physicians’ risk perception and disposition decision making. We conducted a prospective study of adults with syncope in six EDs in four cities over 32 months. We collected patient characteristics, ED management, disposition, physicians’ prediction probabilities at index presentation and followed patients for 30 days for serious outcomes: death, myocardial infarction (MI), arrhythmia, structural heart disease, pulmonary embolism, significant hemorrhage, or procedural interventions. We used descriptive statistics, ROC curves, and regression analyses. We enrolled 3662 patients: mean age 54.3 years, and 12.9 % were hospitalized. Follow-up data were available for 3365 patients (91.9 %) and 345 patients (10.3 %) suffered serious outcomes: 120 (3.6 %) after ED disposition including 48 patients outside the hospital. After accounting for differences in patient case mix, the rates of ED investigations and disposition were significantly different (p < 0.0001) across the four study cities; as were the rates of 30-day serious outcomes (p < 0.0001) and serious outcomes after ED disposition (p = 0.0227). There was poor agreement between physician risk perception and both observed event rates and referral patterns (p < 0.0001). Only 76.7 % (95 % CI 68.1–83.6) of patients with serious outcomes were appropriately referred. There are large and unexplained differences in ED syncope management. Moreover, there is poor agreement between physician risk perception, disposition decision making, and serious outcomes after ED disposition. A valid risk-stratification tool might help standardize ED management and improve disposition decision making.