Industrial machines with gears have been researched from various viewpoints such as efficiency, vibration, and noise. Recently, damage diagnosis from remote administration is demanded as an example of maintenance business. For this reason, a lot of research has dealt with a damage diagnosis on tooth surface of gears by using vibration or noise data. However, these diagnosis methods are difficult to detect early on an abnormal tooth. Therefore, we developed a method to diagnose using a laser beam from the point of view of remote diagnosis. In this paper, we verified our method can accurately early on and diagnose the condition of tooth surface. Our method can be explained as follows: First, a tooth surface is irradiated by a zonal laser beam from oblique direction, and then an irradiated laser beam line is shifted along the height of the tooth according to gear revolution. The variations of laser reflection between the initial and the present conditions are compared, and it can estimate the condition on tooth surface such as initial or abnormal abrasion, pitting, spalling, etc. To confirm the validity of our method, a pitting experiment was carried out, and at the same time the laser reflection was measured as well as vibration and noise. The vibration data of the driving gear and the pedestal as well as the sound data at the adjacent point were analyzed with continuous wavelet transform correspond to each load repetition number. They were compared with the result of the measured data of diffused laser reflection on the tooth surfaces of the drive gear as a function of the rotational angle of the gear. As a result, the laser reflection data revealed occurrence of the pitting more apparently and earlier on than that of vibration and noise. Thus, it can be concluded that our method can accurately estimate the condition of tooth surface and sufficiently assess the duration of the gear's life. Furthermore, to verify that our method can accurately diagnose a tooth surface of a lubricated gear, a measurement experiment in the practical used gearbox was also carried out. By attaching a cover to the laser receiver, the measured data was unaffected by the lubricant. Under the condition of force-feed lubrication, pitting could be detected in every range less than 1800 rpm. Therefore, our method can also diagnose a tooth surface of the lubricated gear in the practical used gearbox. Finally, an automatic damage diagnosis method was developed. Two-stage thresholds were applied to the difference of the diffused laser reflection between damaged data and benchmark data. As a result, this method could detect the occurrence of the pitting, and the value of the first-stage threshold was important to improve the accuracy of the detected length of the damage area.