OBJECTIVES: The first objective of this study was to compare the predictive ability of the GRACE Risk Prediction Index (GRPI) and the Charlson Comorbidity Index (CCI) in determining in-patient mortality, mortality within six months of discharge, and occurrence of a second coronary event or procedure after discharge, in patients who were admitted to a university hospital for the treatment of acute coronary syndrome (ACS).  The second objective was to determine if combining GRPI and CCI in prediction models improved the ability to determine the same outcomes.  METHODS: Data used for this study were obtained from a large academic health system’s acute coronary syndrome patient registry.  Registry data is abstracted from the medical record as well as from patient self-report by a telephone follow-up survey 6 to 12 months after discharge. Logistic regression/ROC was used to derive c-statistics for CCI, GRPI, and CCI-GRPI combined predictive models for each of outcomes.  Likelihood ratio tests were conducted to determine the contribution of CCI when added to predictive models using the GRPI. RESULTS: A total of 1202 patients had complete data from the index hospitalization.  The GRPI model had a higher c-statistic (0.73) versus CCI model (0.68) and similar to the combined model (0.75) to predict in-patient death; the c-statistics for models predicting death during the follow-up period were similar (GRPI=0.74, CCI=0.77, Combined=0.81); and all were similar in predicting secondary events after discharge (GRPI=0.57, CCI=0.60, Combined=0.58).  The Likelihood ratio analysis demonstrated that adding the CCI to existing predictive GRPI models was beneficial primarily for predicting secondary events post discharge.  CONCLUSIONS: The CCI modestly improves models using GRPI to predict death after discharge for an ACS event.  It is an acceptable alternative to the GRPI in predicting death and secondary events if data to derive the GRPI are not available.