Summary: Firstly, the authors describe the stochastic resonance briefly. Secondly, using double-well Langevin equation as an example, they analyze and compare three methods with which the system can turn to stochastic resonance. While output signal to noise ratio (SNR) is used as the measurement of stochastic resonance, tuning system parameter and noise intensity at the same time, the bigger SNR can be gotten than only tuning the system parameter or the noise intensity, and the maximum of SNR is on the lower boundary of noise. Thus, fixing the noise intensity at its lower boundary, and then tuning the system parameter, we can get the same maximum of SNR.