Through the research method of theoretical derivation, quantitative analysis and design of experiment, the relationship among the computational complexity of the mixed-radix Fast Fourier Transform(FFT) algorithm in which N is not the integer power of 2, different combinations and composite sequences of N’s decomposition factors are analyzed. The experimental result indicates, under certain conditions, for the mixed-radix FFT of the same FFT point N, the computational complexity of the mixed-radix FFT algorithm relates to the size of the total K of N’s decomposition factors, but has nothing to do with composite sequences of N’s decomposition factors. An idea is proposed to establish a matching library of the combinations of N’s decomposition factors for mixed-radix FFT, acting as a reference table for obtaining minimal computational complexity when using mixed-radix FFT. It can also provide reference for the related research and engineering personnel in the practical applications.