The Pogo suppression method can be divided into active suppression and passive suppression. For practical engineering, passive suppression is generally achieved through accumulators, but this method sometimes cannot meet the design requirements of liquid rockets with uncertain model parameters and time-varying parameters. As for active suppression methods, state variables are mostly used as feedback thus leading to poor practical feasibility. In this study, an adaptive Pogo active suppression controller design method is proposed where measured acceleration is used as feedback. The eigenspace transformation theory is utilized to design two types of dimensional reduced models: an observation-oriented dimensional reduced model (O-ODRM) and a control-oriented dimensional reduced model(C-ODRM). The former works for observer design, and its state variables can be reconstructed by measuring only the acceleration and can be input to the adaptive controller after matrix operations. The latter is employed for adaptive controller design, and it can effectively process the issues of model parameter uncertainty and time-varying parameters. To demonstrate the effectiveness of the proposed method, a certain type of rocket was selected as the research object for simulation analysis. The simulation results show that the proposed method, which uses measured acceleration as feedback, can achieve Pogo active suppression for liquid rockets with uncertain and time-varying parameters.