Myocardial lipomatous metaplasia, which can serve as substrate for ventricular arrhythmias, is usually composed of regions in which there is an admixture of fat and nonfat tissue. Although dedicated sequences for the detection of fat are available, it would be time‐consuming and burdensome to routinely use these techniques to image the entire heart of all patients as part of a typical cardiac MRI exam. Conventional steady‐state free‐precession (SSFP) cine imaging is insensitive to detecting myocardial regions with partial fatty infiltration. We developed an optimization process for SSFP imaging to set fat signal consistently “out‐of‐phase” with water throughout the heart, so that intramyocardial regions with partial volume fat would be detected as paradoxically dark regions. The optimized SSFP sequence was evaluated using a fat phantom, through simulations, and in 50 consecutive patients undergoing clinical cardiac MRI. Findings were validated using standard Dixon gradient‐recalled‐echo (GRE) imaging as the reference. Phantom studies of test tubes with diverse fat concentrations demonstrated good agreement between measured signal intensity and simulated values calculated using Bloch equations. In patients, a line of signal cancellation at the interface between myocardium and epicardial fat was noted in all cases, confirming that SSFP images were consistently out‐of‐phase throughout the entire heart. Intramyocardial dark regions identified on out‐of‐phase SSFP images were entirely dark throughout in 33 patients (66%) and displayed an India‐ink pattern in 17 (34%). In all cases, dark intramyocardial regions were also seen in the same locations on out‐of‐phase GRE and were absent on in‐phase GRE, confirming that these regions represent areas with partial fat. In conclusion, if appropriately optimized, SSFP cine imaging allows for consistent detection of myocardial fatty metaplasia in patients undergoing routine clinical cardiac MRI without the need for additional image acquisitions using dedicated fat‐specific sequences.