Vibration monitoring consists of the analysis of vibration data coming from vital components of the plant to detect features that reflect the operational state of the machinery. The condition-based maintenance of the bearing leads to the identification of potential failures and their causes, and make it possible to perform efficient preventive maintenance. Early fault detection in bearings is important because it can decrease the probability of catastrophic failures, reduce forces outage, maximize utilization of available assets, increase the life of the plant, and reduce maintenance costs. Subsequent to developing vibration condition assessment procedures that can distinguish between good and faulty bearings, it is necessary to choose requirements for diagnostics and condition prediction. There are four primary activities about bearing fault diagnosis and prognosis. These activities include: detection of changes in bearing condition during operation, short term prediction of bearing serviceability, determination of the type and degree of severity of all life threatening defects appearing as a result of bearing quality, installation, and operation, and prediction of defect development and determination of guaranteed time for trouble-free operation. In this paper, we discuss a newly developed technique for bearing fault diagnosis and prognosis. The proposed techniques is based on a hybrid neural network-fuzzy logic methodology which integrates explicit and implicit expert's knowledge in recognition and classification of bearing faults. Furthermore, we discuss the critical issues regarding the bearing data acquisition, signal filtering, signal pre-processing, and signal features enhancement. Some applications of this techniques for detection of bearing faults of manufacturing machinery are demonstrated. The proposed approach has a number of applications in both manufacturing and aviation maintenance.