The vibration fault diagnosis of hydraulic generator units is researched using the entropy weight coefficient， grey incidence analysis and information fusion technology. The vibration spectrum characteristics and amplitude characteristics are considered as the fault diagnosis samples. An initial vibration fault diagnosis of hydraulic generator units is performed based on the grey incidence analysis and entropy weight coefficient. The final diagnosis results are deduced by fusing different pieces of evidence using the Dempster-Shafter theory. A diagnostic example demonstrates that the proposed method is a good candidate for use as an online diagnosing tool for hydraulic generator units.