Cardiac cine datasets are difficult to obtain in the presence of arrhythmia or poor gating signal. Here, completely ungated radial real-time imaging may help. Cartesian real-time imaging usually offers compromised spatial resolution to maintain sufficient temporal resolution. Hence, we propose a radial bSSFP-based real-time approach with Golden Angle increment which enables view sharing of temporally adjacent projections. To minimize temporal blurring, a narrow tornado shaped filter, followed by subsequent iterative SENSE reconstruction, was used. Remaining streaking artifacts were reduced by a principal component analysis based technique. Results show good agreement in terms of image quality to a standard Cartesian cine dataset.