Gearbox fault diagnosis is mainly based on vibration signal analysis, however, vibration signal acquisition is not easy because of its contact-measuring. Acoustic-based diagnosis (ABD) can overcome this disadvantage. However, for traditional ABD it is hard to choose proper measuring positions and acoustic signal based on single channel measurement and analysis can be used only for local diagnosis. Combining near-field acoustic holography (NAH) with spatial distribution feature extraction, a method of diagnosing gearbox faults based on spatial distribution features of sound field is presented. In view of the corresponding connection between the gearbox running state and its spatial distribution of sound field, firstly, sound field is reconstructed by NAH, and then texture features reflecting distribution characteristics of sound field are extracted from acquired acoustic images. The inner relationship between spatial distributions of sound field and gearbox running states can be established by these features. Finally, combining support vector machine-based pattern classification fault diagnosis can be realized. The effectiveness and superiority of the method is demonstrated on good experimental results.