In this paper a video-based method to automatically track instantaneous velocities of a swimmer is presented. Single cameras were used to follow a marker (LED) attached to the body. The method is inspired by particle tracking techniques, traditionally used in the field of fluid dynamics, to measure local velocities of a fluid flow. During the validation experiment, a white LED was attached to the hip of a swimmer together with a speedometer. A swimmer performed four different stroke types. The velocity profiles using LED tracking were captured and showed less noise than the speedometer measurements. Only at times when the marker disappeared above the water surface due to body role in front crawl and backstroke swimming did the LED tracking fail to capture the athlete’s motion. The algorithm was tested in a 2D case with a single LED to illustrate the proof of principle, but should be suitable for implementation in a 3D analysis or multiple LED analysis.@en