Exercise testing provides valuable information but is rarely integrated to derive a risk prediction model in a referral population. In this study, we assessed the predictive value of conventional cardiovascular risk factors and exercise test parameters in 6,546 consecutive adults referred for exercise testing, who were followed for a period of 8.1 ± 3.7 years for incident myocardial infarction, coronary revascularization, and cardiovascular death. A risk prediction model was developed, and cross-validation of model was performed by splitting the data set into 10 equal random subsets, with model fitting based on 9 of the 10 subsets and testing in of the remaining subset, repeated in all 10 possible ways. The best performing model was chosen based on measurements of model discrimination and stability. A risk score was constructed from the final model, with points assigned for the presence of each predictor based on the regression coefficients. Using both conventional risk factors and exercise test parameters, a total of 9 variables were identified as independent and robust predictors and were included in a risk score. The prognostic ability of this model was compared with that of the Adult Treatment Panel III model using the net reclassification and integrated discrimination index. From the cross-validation results, the c statistic of 0.77 for the final model indicated strong predictive power. In conclusion, we developed, tested, and internally validated a novel risk prediction model using exercise treadmill testing parameters.