In a driving simulator study, driver responses to Steer-By-Wire system failures were investigated. Duration and amplitude of the error were varied, as well as the type of error (with/without torque feedback) and the drivers expectancy of error occurrence. Results are analysed in terms of driver input (steering wheel angle), overall driver-vehicle output (lateral path deviation), and subjective ratings. Drivers base their judgement of error severity on the steering input rather than on the resulting path deviation. Errors with a counter-intuitive torque feedback are rated as worse than errors without torque feedback, even though the overall outcome is better.