Automotive software systems are becoming increasingly complex, driven both by advances in technology and by demands for more powerful applications. The design of such complex systems calls for advanced techniques in requirements engineering. This paper is concerned with developing and validating a new concept in order to elicit and specify user requirements of automotive software functions. Text-based use cases are considered to represent the requirements. The goal of the developed approach is to make the use case creation process more systematic and thereby more efficient (increased quality, reduced time and cost). The concept is based on a systematic expert interview process. After the interview process is completed, use cases are generated automatically. After describing the approach theoretically we show and compare results of an experiment where we applied different parts of the concept in practice.