A new approach for non-blind watermarking of still gray level images in the wavelet domain. The method uses the Human Visual System (HVS) characteristic, and an innovative entropy based approach to create an efficient watermarking scheme is presented in this paper. It decomposes original image in DWT domain in to three hierarchical levels and watermarks it with a logo image, which is scrambled thru a well-known PN-sequence. An entropy-based method is developed for selection of DWT coefficients that provides an adaptive way for determining the number of watermarked coefficients and watermarking factor at each level of DWT decomposition. This approach shows an excellent resistance against almost all the attacks known in the watermarking literature. The detection results of the method reveal better resistance in comparison to the existance methods. With simple modifications, the method can be used for color images and in real time systems.