Malware infections is one of most common security risks which can lead to loss and compromise of data and privacy and often leads to loss of money(Eg Ransomware). The Windows OS is one of most commonly used OS and also most suspectible to malware infections. In our project, we have surveyed and implemented machine learning based technique to classify file as benign or malicious with high accuracy and low computational power. Out of 56 raw features in our dataset, after modification, we have selected only 28 features. We applied various Machine learning algorithms and achieved accuracy upto 99%. Additionally, we analysed false positives and false negatives to un- derstand how file header values control maliciousness of the file. Also, we tried to modify PE header values of few malicious files to check our classifier accurac