The regenerative chatter during milling is a consequence of the machine tool dynamic compliance and the specific setting of the cutting process. The basic request on the machine structure is to enable a stable cut with dominant portion of the installed spindle power. In order to improve the dynamic properties of newly-designed machine tools, the machine tool structures are optimized using various advanced simulation methods and are made of various advanced materials. The final design should be tested experimentally. The test of usable spindle power provides important feedback on the machine tool design. The testing of large machine tools is time consuming due to the significantly varying machine dynamic compliance dependent on the working space. Due to the installed high power and large diameter tools used, the occurrence of chatter might be dangerous with a high risk of machine tool damage. The article presents a method for the safe and time-effective experimental testing of the machine tool usable spindle power. The method is based on the experimental identification of the cutting force coefficients and the experimental identification of machine tool behavior with the support of extrapolation models based on identified experimental data. The dynamic compliance is measured within a rough network of measurement points in the machine working space. The measured dynamic compliance results are characterized by a model based on modal identification data. Only a few cutting tests are done for the cutting force coefficient identification. Finally, a map of the usable spindle power is calculated. This approach makes it possible to calculate the results with respect to the real machining process and the current machine tool conditions across the whole working space in an acceptable time. The approach is demonstrated on a real case study.