Breast cancer is the top cancer in women both in the developed and the developing world. Unfortunately, most of breast cancer is diagnosed in very late stages. Present there is no known method to prevent breast cancer but early detection increase the chance of cure. Developing computer aided diagnosis system (CAD) can help radiologist in their decision. In this paper, we proposed a CAD system to segment and classify the breast cancer in ultrasound images. The system is using marker controlled watershed transformation technique to identify the region of interest (ROI). Then, wavelet transform is applied to extract set of features combined with texture and statistical features. The classification step determines whether the ROI is normal or focal lesion. Finally, focal lesion is classified as benign or malignant. Support vector machine (SVM), K-nearest neighbour (KNN) and classification & regression trees (CART) are used to achieve the classification task. The proposed method is validated using 10 folds cross validation and the obtained results were encouraging. The results show that CART obtain 83.75% classification rate using statistical and texture features in case of classify benign and malignant tumor which is more than SVM and KNN classification rate. In case of differentiate between normal and abnormal classes SVM and CART obtain 100% classification rate using texture feature.