A vertical alignment liquid crystal display includes two sub-pixels each with a variable capacitor. A pixel is bisected into a high gray sub-pixel and a low gray sub-pixel through forming a variable capacitor at each sub-pixel. With this structure, the sub-pixels express different grays so that lateral visibility is enhanced. It is not required in bisecting a pixel into two sub-pixels to form separate wires for applying different signals thereto, and the amount of data to be processed at the driver for driving the display device is reduced. Furthermore, a pixel is bisected into two sub-pixels with variable capacitors in a simplified manner, and it is not required to form additional wires and elements, so the aperture ratio is enhanced.