Objective. The soluble form of ST2 (sST2) is a novel laboratory parameter for cardiac risk prediction, and over the past years, several studies have tried to evaluate its utility, especially in the management of heart failure. We investigated whether increased serum levels of sST2 show a characteristic pathomorphologic pattern in 3-Tesla cardiac magnetic resonance imaging (CMRI). Methods. One hundred and fifty-six patients referred to 3T CMRI due to suspected coronary artery disease (CAD) or myocarditis were prospectively enrolled in the study. Ninety patients were diagnosed with CAD, 22 patients with myocarditis, and 44 patients, who constituted the reference group, showed no pathologic CMRI pattern. Results. There was no significant difference between the sST2 values for patients in the reference group and patients with CAD or myocarditis. The sST2 concentration showed a weak correlation with the NYHA functional class (P = 0.002, r = 0.22), but correlation of sST2 and LGE, left ventricular parameters, and LVEF could not be seen. In contrast NT-proBNP was positively correlated to left ventricular parameters, LGE, and NYHA class function (P < 0.05). Additionally, it showed an inverse relationship to LVEF (P < 0.001, r = - 0.42). Conclusions. Soluble ST2 is not able to detect myocardial scar and should not be used alone as a parameter for detection of inflammation and myocardial scar formation.