Rolling element bearing are key mechanical parts in rotating machinery and large equipment such as helicopters, wind turbines and ships. Running under severe working environments and suffering extreme load and speed conditions, bearings may be damaged. Bearing fault will degrades the overall performance of the transmission system, thus monitoring and fault detection of bearing is necessary for industry. However, bearing monitoring signal is usually influenced by noises, and traditional spectrum and time-frequency analysis cannot reflect the phase information within the signal. To address this problem, a fault detection method based on modulated signal bi-spectrum is proposed in this paper after reexamining characteristic of bearing monitoring signal, the method is validated using experiment signals. The result indicates that the method proposed in this paper is effective to detect bearing fault in transmission drive train.