The study applied a retrospective follow-up design to determine the prognostic effect of graded exercise testing (GXT) in patients with low- to moderate-risk chest pain evaluated in an emergency department 9-hour protocol chest pain center (CPC) from January 1, 1993 to August 1, 1996. The cohort of 1,209 patients were followed to the date of death or first adverse cardiac event up to 1 year after CPC admission. Cardiac events were defined as coronary artery bypass graft, percutaneous transluminal coronary angioplasty, cardiogenic shock, cardiac-related death, congestive heart failure admission, ventricular tachycardia/ventricular fibrillation arrest, and myocardial infarction. Patients with acute ST-segment elevation or depression of >1 mm, positive enzyme (creatine kinase myocardial band) testing, or unstable angina during their CPC evaluation were admitted without GXT testing. Statistical analysis included chi-square test for complication rates and Cox proportional-hazards modeling. Nine hundred fifty-eight of 1,209 patients underwent GXT testing. Patients with positive, inconclusive, and normal GXTs had complication rates of 36.8% (7 of 19), 3.4% (9 of 267), and 1.1% (5 of 456), respectively. After adjusting for age, sex, and race, the relative risk of complication was 38.9 (95% confidence interval 11.7 to 129.6) with a positive GXT, and 3.6 (95% confidence interval 1.2 to 10.7) with an inconclusive GXT compared with a normal GXT. The GXT is a good prognostic indicator of adverse cardiac events in low- to moderate-risk chest pain in patients evaluated in an emergency department CPC.