Acute coronary syndrome registries report the use of incomplete guideline therapies, especially among the highest risk patients. Whether this treatment gap results from misperceptions of risk by physicians is uncertain. The Perceived Risk of Ischemic and Bleeding Events in Acute Coronary Syndrome Patients (PREDICT) study was a prospective acute coronary syndrome registry in Australia, China, India, and Russia, involving 58 hospitals between May 2009 and February 2011. In-hospital care and events up to 6 months were assessed. At least 2 clinicians involved in patient care estimated the untreated risk and change in risk with each therapy. Physician risk assessment and objective risk measures (eg, Global Registry of Acute Coronary Events [GRACE] score) for death, death/myocardial infarction, and bleeding events were compared using the c statistic and integrated discrimination improvement. In total, 1542 patients and 4230 patient-specific physician estimates were obtained. Of responding clinicians, 81.9% were cardiovascular specialists (years of practice: mean [SD], 11.5 [7.7] years). The median physician-perceived risk of 6-month death was 25% (interquartile range, 14%-35%). The GRACE score was superior to physician estimation (c statistic: GRACE score, 0.812 [95% confidence interval, 0.772-0.851] versus physician, 0.652 [95% confidence interval, 0.596-0.708]; P<0.0001). The GRACE score added to clinician intuition improved discrimination (integrated discrimination improvement, 0.0632 [SE, 0.012]; P<0.0001). Invasive management correlated with physician-estimated risk but not with GRACE risk. Among patients not at high risk by physician estimation, increased risk by GRACE score was associated with higher mortality (3.7% versus 0.8%; P<0.001). Objective risk assessment provides superior risk discrimination when compared with physician-estimated risk. Whether systematic use of objective risk stratification improves clinical outcomes should be studied in appropriately designed clinical trials.