Abstract: Giving a presentation is an everyday skill in many people’s educational and professional life. However, training is still rare and expensive. Important aspects like talking speed or body language are well understood, and many good practices exist, but they are difficult to evaluate. They require at least one experienced trainer who attends the presentation and is able to evaluate and give a constructive feedback. Our aim is to make a first step towards an automatic feedback system for presentation skills by using common motion-detection technology. We implemented a software tool using Microsoft’s Kinect and captured gestures, eye-contact, movement, speech, and the speed of slide changes. A short evaluation using eight presentations in a university context showed that speaker movement and body gestures are detected well while not all spoken words and slide changes could be recognized due to the Kinect’s limited technical capabilities.