There are different types of tumors are available. Astrocytoma is the most common type of tumor (30% of all brain tumor) and is usually a malignant one. Astrocytoma can be subdivided into four grades. Each grade has its own characteristics and unique treatment. If any wrong treatment is given to these grades that leads to death. So finding the position and shape of tumor is very important for the further treatment. The proposed system of this paper is to find the exact position and shape of the tumor cells. That helps the physician for further treatment. In the proposed system, it consists of four modules (i) Pre-processing, (ii) Segmentation of brain in MR Images,(iii) Quality extraction and (iv) Approximate reasoning. Pre processing is done by filtering. Segmentation is done by advanced K-means and Fuzzy C-means algorithms. Quality extraction is by thresholding. Finally, Approximate reasoning method to recognize the tumor shape and position in MRI image. If the tumor is a mass in shape then k-means algorithm is enough to extract it from brain cells. Suppose if it is a malignant (spread over the brain) one then the Fuzzy C-means algorithm will be used for accurate tumor diagnosis, since the Fuzzy method is used for floating point prediction of the tumor cells. At the end of the process the tumor shape, position, area and its stage will be determined. In this project the two strong algorithms are used for segmentation. So, the entire system for tumor segmentation is more accurate than other methods.