When the rolling bearing of the gearbox failed， the fault pulse is annihilated by the strong interference signals， which makes it difficult to extract the fault characteristics. To solve this dilemma， a fault diagnosis method of rolling bearing based on pulse enhancement extraction is proposed. Based on the envelope spectrum that characterized the periodicity of the fault pulse， a Pulse Extraction Operator index （PEO） is constructed to evaluate the strength of the fault pulse in the vibration signal. Taking the need for enhancement of the weak fault signal features into account， a Pulse Enhancement Extraction Operator index （PEEO） is constructed combined with the Minimum Entropy Deconvolution （MED）， which can evaluate the effect of different MED filter lengths on fault pulse enhancement. Then a three-dimensional filter model based on MED filter length， pulse frequency and PEEO is formed. With the PEEO peak value， the optimal filter length of the MED and the characteristic frequency of the pulse-enhanced signal can be located， and the fault features of gearbox rolling bearing can be extracted while obtaining the optimal pulse enhancement signal. Simulation analysis and experimental verification results show that this method can effectively enhance the pulse of bearing fault， and present significant bearing fault characteristics in the PEEO amplitude spectrum of the optimal pulse enhancement signal， so as to realize the diagnosis of weak fault of gearbox rolling bearing. The comparison also shows the advantages.