This paper discusses the method to automatic mass segmentation and analysis of mammogram for classification of benign or malignant tumor. The identification process is started with image enhancement through cropping to remove artifacts, then followed by increasing the contrast through Contrast Limited Adaptive Histogram Equalization (CLAHE). The selection of mass candidate is carried out through 2 phases: marking the suspected mass area (Region of Interest - ROI) using adaptive thresholding p-tile technique and marking the connected components, and texture feature extracting on the ROI to classify whether the ROI is mass or non-mass. The texture feature extraction is performed by Grey Level Co-occurrence Matrices (GLCM) set up on four different directions, 0°, 45°, 90°, and 135°. The application captures a mammogram image as an input and displays the presence of suspicious mass and its margin, if any. The segmented mass is analyzed based on its shape and margin. Thereafter, these information can be used by physicians to classify the type of tumor and to decide whether a biopsy is necessary. The application is evaluated using the mammogram data from Mammographic Image Analysis Society (MIAS). The MIAS data consist of 207 images of normal breast, 64 benign, and 51 malignant. 85 mammograms of MIAS data have mass. It is tested using Mammogram from Picture Archive Communication System (PACS) Pertamina hospital. Based on the study conducted, the algorithm developed step by step can localize the suspected area therefore it is able to detect the shape and the edge of mass on mammogram.