Abstract: The brain tumors are most common and aggressive disease, leading to very short life expectancy rate. Thus, treatment at initial stage is a key stage to improve the quality life of patients. There are image techniques such as Computed Tomography (CT), Magnetic Resonance Imaging (MRI) , ultrasound image which are used to evaluate the tumors in brain, lung and liver. MRI images are used to diagnose tumors in the brain. There are few limitations for the manual classification of tumors. Hence automatic classification scheme is introduced to prevent the death rate of human. In this work, automatic classification is done through MR images using neural networks. In our proposing system, we are implementing new algorithm for detection of tumors with more accuracy. Thus, the automatic tumor detection will more accurate.