Purpose: To optimize and characterize an interrupted 5D free-running framework at 3 T for detailed cardiac function assessment without the use of breath holding or contrast agents. Methods: A free-running 3D radial gradient echo sequence was periodically interrupted with a T2 preparation and a recovery module to optimize native blood-to-myocardium contrast at 3 T. Lipid signal was suppressed using a numerically optimized water-excitation RF pulse to reduce lipid streaking artifacts and to improve overall image quality. Optimal acquisition parameters were established for a 5-min scan time using extended phase graph simulations. A compressed sensing-based reconstruction incorporating cardiac and respiratory inter-bin deformation fields was employed to generate 5D images of the whole heart. The sharpness and contrast between the left ventricular blood pool and myocardium, along with the functional measurements of the left ventricle from the 5D datasets, were compared to routine 2D cine imaging in 16 healthy volunteers and three patients referred for clinically indicated CMR. Results: The proposed method resulted in lower contrast (0.57 +/- 0.12 vs. 2.09 +/- 0.74,p <0.001) and sharpness (3.76 +/- 1.11 mm vs. 2.74 +/- 0.95 mm, p <0.001), but enabled similar left-ventricle ejection fraction assessment (bias = 1.3%, limits of agreement = [-3.3%, 5.9%], intraclass correlation coefficient = 0.87,p = 0.03) with high reproducibility compared to 2D cine. Conclusion: The proposed contrast-free, interrupted free-running 5D imaging provides left ventricular functional assessments comparable to 2D cine at 3 T, while offering an improved patient experience through shorter scan times and free breathing.