Seismic simulators are designed to recreate the effects of seismic activity on structural and geotechnical systems. The performance objective in recreating the effects of a particular ground motion on a test article is typically for the seismic simulator to track the pre-recorded motion of a particular historical earthquake. The reference signal is often the earthquake acceleration, as the dynamic loads imparted upon the structure are proportional to this acceleration. During seismic simulations, interactions between the test article and the actuators may be significant; the inertia forces of the test article may not be much less than the actuator force capacity. In such cases, the table motion is not an exogenous input to the test article, but is an aspect of the response of these two coupled systems. This study examines the suitability of actuator acceleration tracking accuracy as the performance objective for seismic simulators by examining the dynamics of linearized models for coupled systems of actuators and test articles. A test metric related to the test system inertia and the test article inertia indicates the degree to which accurate accelogram tracking may not recreate the desired prototype responses. © 2009 AACC.