One of difficulties to detect abrasive wear in journal bearings through vibration analysis is the lack of obvious fault characteristics in an early stage. To carry out an in-depth investigation on the relationship between wear occurred inside and vibration signals acquired outside is meaningful from the diagnostic point of view. In this paper, a tribo-dynamic model is developed for the fluid-asperity interaction in journal bearings. The microscopic pressure fluctuation induced random excitation is derived. The effects of surface roughness and operating condition on the random excitation are investigated. It is found that wear induced narrowband spatial components of journal surface can excite random vibration of bearing. The speed dependent vibrational behaviour is found to be an effective indicator of surface defects.