Bearing is one of vital transmission elements, finding numerous applications in small, medium and large machineries. Sound and vibration signals of a rotating machine contain the dynamic information about its operating condition. There are many articles in the literature reporting suitability of vibration signals for fault diagnosis applications; however, the transducer used for vibration signals (accelerometers) is costly. This prevents small scale industries and low cost equipments from using diagnostic tools on affordability ground. On the other hand, transducers used for acquiring sound signals are relatively low cost or/and affordable. Hence, there is a need for studying the use of sound signal for fault diagnosis applications. This paper uses acoustic signals (sound) acquired from near field area of bearings in good and simulated faulty conditions for the purpose of fault diagnosis through machine learning approach. The descriptive statistical features were extracted from sound signals and the important ones were selected using decision tree (dimensionality reduction). The selected features were then used for classification using C4.5 decision tree algorithm. The paper also discusses the effect of features, effect of various classifier parameters on classification accuracy. © 2012 Elsevier Ltd. All rights reserved.