Machine tool models play an important role in supporting decision-making for machine tool procurement, process planning, and production scheduling in manufacturing. Due to modeling uncertainties, however, it is challenging to create a machine model that accurately and dynamically represents the real machine. Modeling uncertainties and errors can be introduced during the model development process (i.e., when a machine model is created from scratch) and the model conversion process (i.e., when a machine model in one format needs to be converted into another format, e.g., from a vendorspecific format to a neutral format such as the STandard Exchange of Product Data (STEP)). This paper identifies these uncertainties and provides a methodology to help ensure correct conversion of the coordinate system from one definition to another. Examples are provided to illustrate the methodology. As a fundamental method, this methodology will help improve model accuracy by compensating multiple modeling errors. It also supports the synchronization between machine models with real machines when building digital twins of the machine tools by adjusting the coordinate system offsets continuously. Digital twins of machine tools would be an integrated solution for addressing most of the modeling uncertainties in order to constantly monitor the status of the machine tool, dynamically update the model parameters, and in turn optimally control the machine tool.