Friction stir welding (FSW) is a new kind of solid state welding technique. Rigid and reliable joints in intricate shapes are possible with FSW. Mutual transfer of material occurs between two work pieces when heat is generated by continuous stirring of the welding tool. This type of welding process is frequently used in many commercial applications like automobile, ship building, aerospace and many more. In this scenario, monitoring the FSW tool condition is essential in order to avoid the early defects and breakdown of the machine. The FSW tool condition monitoring offers numerous benefits in the fabrication of aluminium products with less weld defects. Condition monitoring of friction stir welding tool is an advanced predictive maintenance technique for collecting real time data from the operating machine through sensors. The collected data can be analyzed using a machine learning approaches. In this study Al alloy was used for experimentation by using vibration analysis techniques signals were captured for good and faulty conditions of the tool. Statistical information was extracted from the raw vibration signatures and selection of feature was carried out. The selected features were then classified using Best first tree (BFT) classifier. The post pruned best first tree produced 93.07% as the classification accuracy. © 2019 Elsevier Ltd.