To derive and validate a prediction rule in patients with acute chest pain (CP) without existing known coronary disease.Cohort study including 2233 patients with CP. Based on clinical judgment, 1435 were discharged as very low risk and the remaining 798 underwent exercise tolerance test (ETT). End point: 6-month composite of cardiovascular death, nonfatal myocardial infarction, and revascularization. The prediction rule was derived from a randomly selected test cohort (n = 1106) summing factors of variables selected by multivariate regression analysis: CP score higher than 6 (factor of 3), male gender, age older than 50 years, metabolic syndrome, and diabetes mellitus (factor of 1, for each). The prediction rule was validated in the remaining cohort (n = 1127). All patients with CP were categorized into 3 groups: group A (prediction rule 0-1), B (2-4), or C (5-6). Outcomes and prognostic yield of ETT were compared among each group.In the test cohort, 55 patients (5%) reached the composite end point. Event rate increased as the prediction rule increased: 1% for group A, 6% for B, and 25% for C (P < .001). This pattern was confirmed in the validation cohort (P < .001). A normal ETT did not significantly improve the high (99%) negative predictive value in group A and did not succeed in excluding the composite end point (17%) in group C.In patients with acute CP without existing coronary disease, a prediction rule based on clinical characteristics provided a useful method for prognostication with possible implication in decision making.