For diagnostics and therapy planning, the segmentation of medical images is an important pre-processing step. Currently, manual segmentation tools are most common in clinical routine. Because the work is very time-consuming, there is a large interest in tools assisting the physician. Most of the known segmentation techniques suffer from an inadequate user interface, which prevents their use in a clinical environment. The segmentation of medical images is very difficult. A promising method to overcome difficulties such as imaging artifacts are active contour models. In order to enhance the clinical usability, we propose a user-driven segmentation approach. Following this way, we developed a new segmentation method, which we call interactive snakes. Thereto, we elaborated an interaction style which is more intuitive to the clinical user and derived a new active contour model. The segmentation method provides a very tight coupling with the user. The physician is interactively attaching boundary markers to the image, whereby he is able to bring in his knowledge. At the same time, the segmentation is updated in real-time. Interactive snakes are a comprehensible segmentation method for the clinical use. It is reasonable to employ them both as a core tool and as an editing tool for incorrect results.