A new definition of fuzzy entropy of interval-valued fuzzy sets is proposed and applied to the image thresholding segmentation.A fuzzy parameter is determined according the histogram of image.The image is represented by an interval-valued fuzzy set through selecting an appropriate primary membership function.Then a new method of image thresholding segmentation based on fuzzy entropy of interval-valued fuzzy sets is proposed.Numerical examples show the rationality and practicality of the new method in threshloding segementation of images with noises.