Rolling element bearings are the essential components of rotating machines, faults of which can cause serious failures or even major breakdowns of a machine. Fault diagnosis deliveries significant benefits to machines with rolling element bearings by finding the faults at early period and taking corrective actions to enhance safe and high performance operations. However, multiple sensor usages and high rate data acquisition involved in a monitoring system have considerable drawbacks of high system cost involved in purchasing hardware for data transfer, storage and processing. To reduce these shortages, this paper investigates compressive sensing (CS) techniques for the fault detection of rolling element bearings. Based on the frequency shift and envelope analysis, a CS scheme is developed for monitoring the bearing. The number of data transmitted and stored can be reduced by several thousands of times. The simulation and the experimental results demonstrate that the compressed vibration signals of rolling element bearings are effective to detect bearing faults at the total compressing ratio up to several thousand with the corresponding maximum compression ratio (CR) of CS process at nearly 100. In addition, several performance measures are applied to evaluate the reconstructed signals and show approximately the information about the noise level of the system.