The invention, which relates to the field of communication signal processing, discloses a feedback-structure-based multi-harmonic signal under-sampling method and system. The system is formed by two series sampling channels including a main sampling channel and a feedback sampling channel. For a multi-harmonic signals with K frequency components, the main sampling channel obtains 2K samples and estimates a possible solution combination of a frequency parameter and an amplitude parameter based on a Prony algorithm; according to an estimation result, a sampling rate of the feedback channel is generated; and after the feedback sampling channel obtains 2K samples, a possible solution combination of a frequency parameter and an amplitude parameter is estimated based on the Prony algorithm; andwith the estimation results of the two channels, a frequency parameter and an amplitude parameter of a signal are determined jointly. According to the invention, frequency aliasing and mirror image frequency aliasing problems can be solved by only 4K samples; and the parameter estimation accuracy is high and the robustness is strong.