Simulating a driving environment that provides realistic engine sounds as auditory feedback to the driver can improve the immersion in a driving simulator. However, this is challenging due to the complexity of engine sounds and a wide range of vehicle speeds. This paper describes a method of engine sound simulation and generation for a driving simulator, using Spectral Modeling Synthesis (SMS) to decompose the sound samples into deterministic and stochastic components, and then to resynthesize the sounds with these two components. A linear interpolation method is proposed to generate engine sounds not available in the database from the sounds recorded at two nearby engine speeds. The method is successfully implemented in our driving simulator. Overall, the proposed method provides a practical solution for generating engine sounds of satisfactory quality for driving simulations.