This article studied the use of stochastic resonance of the under-damped second-order linear system to reduce the multiplicative noise. Research shows that when intrinsic frequency and input signal of the system are subject to multiplicative Gaussian white noise, under-damped second-order linear system shows the phenomenon of stochastic resonance. The average output amplitude gain (OAG) is non-linear and under certain conditions the value is greater than that of no noise second-order linear system, and proper adjustment of the system parameters and noise intensity can improve the output amplitude gain. So as long as the system is in the resonance region, weak signal detection of under-damped second-order linear systems can be fulfiled in the noise background. The article used visual simulation software SIMULINK to simulate some examples. Simulation confirmed that this method resolveds multiplicative noise feasibly and effectively.