Conventional compensation of position errors of machine tools relies only on measured values. Due to this principle it is not always possible to compensate the errors in time, especially dynamic ones. Moreover, the relevant control variables cannot always be measured directly. Thus, this approach proves to be insufficient for high precision applications. In this context, a model-based error prediction allows for minimal position errors. However, ultra-precision applications set high demands for the models' accuracy. This paper presents the design of an accurate and real time-capable structural model of an ultra-precision positioning system. The modeling method for the developed ultra-precision demonstrator is shown and the initial parameter identification is presented. © 2017 The Authors. Published by Elsevier B.V.