To distinct among various brain pathologies the information output from spectroscopy is one of the very important pointers. The attitude is to examine the effectiveness of a pattern recognition system in the task of discerning between meningiomas and metastatic brain cancers. The information used for this analysis was a set of Magnetic Resonance Spectroscopy images copied from internet with verified tumors. In this analysis we have applied Discrete Wavelet Transform for the attribute extraction and Principal Component Analysis algorithm for attribute reduction and Support Vector Machine for classification. The system was accomplished using an external cross verification process.