MRI is used to detect other unusual features of brain tumors. The cancer appears when the cell gets damaged. It can appear anywhere in the brain with various sizes and shapes. They can have long-term effects and psychological impacts on brain health and overall life of patient. The evaluation of brain tumors using imaging techniques is becoming more prevalent. Due to differences in the intensity of the images in the MRI images, the results of automated analysis can be inaccurate. The proposed method uses simple techniques of brain tumor segmentation, detection and classification, which differentiates MR images into malignant and benign tumors. The methodology uses Otsu thresholding and Lloyds clustering followed by morphological filtering to segment brain MR images and uses discrete wavelet transform to extract wavelet features from segmented MR image. The probabilistic neural network has been used for classification. Finally radial basis function network performs classification by measuring the inputs vector similarity with prototype vector of training set The current methodology would be useful in clinical practice for the detection of brain tumor.