In order for individuals suffering from transfemoral amputation to walk in a variety of circumstances, the above-knee prosthesis based on posture recognition was presented. The body posture of lower limb was classified into four classes, "stair", "sitting", "standing", and "walking". For measure the amputee's movement intent, surface EMG signals which can reflect amputee's movement intent and can be measured without invasion were applied to identify postural adjustments by support vector machine. The result of this study indicates that this method can recognize every postural adjustment with a higher identification rate, and has a great potential in practical application of artificial lower limb.