Start-up energy consumption is a vital component of machine tool energy consumption. In some cases, such as forecasting and quota planning for energy consumption, the start-up energy consumption must be acquired before practical machining, so that the determination of start-up energy consumption before machining is indispensable. However, the start-up energy is usually neglected or regarded as a part of the no-load energy because its characteristics are complicated and unavailable by theoretical computing models. These treatments could result in significant errors under certain working conditions and could also undermine increasing the accuracy requirements for energy analysis. To close this gap, this article presents a method for determining start-up energy consumption based on specific databases consisting of data and functions. To use the database, the relative change in the rate of power is utilized to identify a start-up process, and a relation model between the start-up energy and the spindle speed is established. The energy consumed in the starting process is obtained by summing the energy consumption of each recording interval or by integrating the power function in its time domain. This experimental study proves that this approach has a relatively high accuracy. In addition, the analysis of its application shows that this method is practical for forecasting energy consumption of a start-up process, planning the energy quota for machining a piece of work in manufacturing and other purposes.