Background Mesenchymal stem cells (MSCs) transplantation provides a new approach for myocardial repair. However, many important fundamental questions about MSCs transplantation remain unanswered. There is an urgent need to identify MSCs from the beating heart and analyze the efficacy of this new approach. This study aimed to localize the magnetically labeled MSCs (MR-MSCs) and monitor the restorative effects of MR-MSCs with magnetic resonance (MR) imaging. Methods Acute myocardial infarction (AMI) was created in swine by a balloon occlusion of the left anterior descending coronary artery. Cells were delivered via intracoronary infusion after myocardial infarction. Infarct size change and cardiac function were assessed with 3.0T MR scanner. The results were then confirmed by histological and western blot analysis. All statistical procedures were performed with Systat (SPSS version 12.01). Results A total of 26 swine were divided into four groups (sham-operated group, n=6; AMI group with PBS transplantation, n=6; labeled MSCs group, n=7; unlabeled MSCs group, n=7). MSCs, MR-MSCs (107cells) or PBS were delivered by intracoronary injection after MI and serial cardiac MR imaging studies were performed at 0, 4 and 8 weeks after transplantation. MR imaging demonstrated MI size decreased after MSCs transplantation in labeled and unlabeled groups, however, increases were seen in the AMI group at 8 weeks after MI. The left ventricular ejection fraction (LVEF) was slightly increased in the AMI group ((41.87±2.45)% vs (39.04±2.80)%, P＞0.05), but significantly improved in the MR-MSCs group ((56.85±1.29)% vs (40.67±2.00)%, P＜0.05) and unlabeled group ((55.38±1.07)% vs (41.78±2.08)%, P＜0.05) at 8 weeks after treatment. MR-MSCs were further confirmed by Prussian blue and immunofluorescent staining. Western blot analysis demonstrated that there was an increased expression of cardiomyocyte markers such as myosin heavy chain and troponin T in the MSCs treatment groups and the ratio of matrix metalloproteinase 2 to tissue inhibitor of metalloproteinase 1 decreased in the labeled group and unlabeled group compared with the AMI group and sham-operated group. Conclusion Transplanted MR-MSCs can regenerate new myocardium and prevent remolding in an MI model at 2-month follow-up and represent a preferred method to better understand the mechanisms of stem cell therapy in future clinical studies.