Understanding the viability of the ischemic myocardial tissue is a critical issue in determining the appropriate surgical procedure for patients with chronic heart failure after myocardial infarction (MI). Conventional MI evaluation methods are; however, preoperatively performed and/or give an indirect information of myocardial viability such as shape, color, and blood flow. In this study, we realize the evaluation of MI in patients undergoing cardiac surgery by Raman spectroscopy under label-free conditions, which is based on intrinsic molecular constituents related to myocardial viability. We identify key signatures of Raman spectra for the evaluation of myocardial viability by evaluating the infarct border zone myocardium that were excised from five patients under surgical ventricular restoration. We also obtain a prediction model to differentiate the infarcted myocardium from the non-infarcted myocardium by applying partial least squares regression-discriminant analysis (PLS-DA) to the Raman spectra. Our prediction model enables identification of the infarcted tissues and the non-infarcted tissues with sensitivities of 99.98% and 99.92%, respectively. Furthermore, the prediction model of the Raman images of the infarct border zone enabled us to visualize boundaries between these distinct regions. Our novel application of Raman spectroscopy to the human heart would be a useful means for the detection of myocardial viability during surgery.