In the development of machining center spindles, electric power consumption was measured and evaluated by use of the S/N ratio. When the S/N ratio was obtained, the ideal state was taken to be a proportional relation between rotational speed and electric power, and the evaluation was carried out using dynamic characteristics. However, from the results of actual electric power consumption measurements, it became clear that the relation is nonlinear for some types of spindle motors. The evaluation method was therefore changed to a method using the standard S/N ratio, and variability and linearity were evaluated separately. Experiments showed less temperature rise under the improved conditions than before the improvement, demonstrating that temperature can be decreased by electric power evaluation. Use of the standard S/N ratio also demonstrated the effectiveness of separating variability and linearity.