Model-in-the-loop (MiL) testing refers to the scenario in which the test specimen is part real and part virtual, i.e. a physical sub-system is linked to a real-time computer simulation. Due to imperfect actuator behavior the hybrid test specimen will not respond in quite the same way as the real (complete) specimen would. In this paper the accuracy of the MiL system is assessed for two automotive testing examples: tire vertical dynamic emulation, and the emulation of aerodynamic forces. Linear models of racecar dynamics are used in the assessment, combined with real measurements of actuator response. The results indicate that considerable care is required in designing such test systems; the tire emulation results in particular are unacceptable.