In this paper we proposed a new method for edge detection in ultrasound images by using a fuzzy inference system (FIS). The input image is preprocessed and the difference in intensities of each pixel to its neighborhood is given as input to the FIS. Depending on the magnitude of the differences, sufficient inference rules were defined in the system to get the edges even in low contrast region. This method is successfully applied to the echocardiogram images to detect the left ventricle (LV) endocardial borders and some experiments are given.