The stability of a machining process is a function of the dynamic response between the spindle and table, which varies within the machine work volume. This paper deals with computationally efficient methodology to evaluate and simulate dynamic performance of the machine tool. A position-dependent virtual model is assembled using finite element model reduced via component mode synthesis and transformed to a state-space multi-input-multi-output system. Combination of these techniques allow time-efficient response simulations with significantly less computational effort than conventionally used full finite element models. The presented approach can be used to create position-dependent dynamic stiffness map within the work volume used to predict and reduce unstable behaviour during operation. Furthermore, these techniques are not reserved for machine tools exclusively and can be used in wider spectrum of technical applications, that require time-efficient response simulations.