Epicyclic geartrains are often preferred in heavy-duty machinery owing to their abilities such as transmitting large amounts of power with minimal loss, good load sharing capacity, large reduction ratios, and compact design. Machinery employing such complex geartrains need an effective monitoring system to predict gear failure at an early stage which prevents catastrophic failure. In this work, vibration signal of the geartrain is acquired using an accelerometer under various gear fault conditions such as healthy gear, defect in sun gear, defect in planet gear, defect in ring gear, defect in both sun and planet gears respectively. Then, statistical characteristics or features such as mean, median, mode, variance, skewness, kurtosis, standard error, standard deviation, maximum and minimum, of the time domain vibration signals are extracted. Afterward, a decision tree algorithm is used to select the most useful statistical features. These selected features form input to the fuzzy classifier. The fuzzy classifier is then modelled to classify the faults based on the rules generated by the decision tree. The classifier output obtained was found to have higher prediction accuracy which signifies the reliability of the proposed method.