Self-driving cars are announced to be available at the end of this decade. For testing these highly complex systems, experiments on proving grounds and long-term validations on public roads need to be complemented by extensive tests in virtual environments. However, efficient planning of these virtual tests is essential to focus only on those tests, which are affected by changes to the software during the development. In previous work, meta-data from the development process of a real-scale self-driving car was analyzed to develop an approach, which relates changes in the source code to scenarios for virtual testing environments. In this article, results from applying that approach during the development of a lane-following algorithm for a self-driving miniature car are presented. With the help of the approach, 11% of the tests in the virtual environment could be omitted safely to save virtual testing time.