Gear inspection equipment is expensive, test procedures with contact-type sensors are time consuming, and most methods require the use of special hardware (such as a master gear) for each gear geometry. In this paper, use of a machine vision system is proposed to detect major gear manufacturing defects. The system obtains the digital image of a single tooth with a video camera and frame grabber. The tooth profile is obtained with the proposed Fast Tooth Profile Detection (FTPD) algorithm, which scans only a small portion of the digital image by considering tooth geometry. The profile error is found by calculating the difference between the inspected tooth profile and a computer generated involute curve. The proposed system obtained a 9-micrometer resolution when a gear with a 4.469″ (113.5126 mm) outer diameter and 50 teeth was inspected. The FTPD algorithm was found to be approximately 20 times faster than the commonly used edge detection algorithm (the Sobel operator) in gear inspection. Also, the FTPD algorithm eliminated the false edge detection problems of the Sobel operator around shadows and bright points. The proposed system is very useful and convenient for detecting major gear defects in less than one minute.