Segmentation of the left ventricle (LV) is a hot topic in cardiac magnetic resonance (MR) images analysis and still remains challenging. In this paper, we propose a novel method, radial gradient vector flow (GVF) snake, to segment LV automatically. Left ventricle centroid and region of interest (ROI) are located using temporal intensity difference along with Hough transform. Taking the centroid as an origin, the ROI could be transformed into polar coordinates where myocardium looks more like a horizontal band rather than a circle. This shape characteristic enables snake to evolve towards 1D radial direction instead of 2D image plane, which simplifies snake energy functions to 1D. A line-like shape constraint is adopted to conquer papillary muscle and weak boundaries. After endocardium extraction, GVF external force modified skillfully reactivates snake forward to epicardial contour successively. Several experiments are presented to demonstrate the effectiveness and robustness of the proposed method.