For the control of uncertainty chaos systems, a compensation control method using the polynomial-basis-functions model (PBFM) based on conjugate gradient algorithm (CGA) is studied. In the proposed method, dynamic properties of a chaotic system are first fitted by PBFM, and then feedforward compensation control for the uncertainty chaos system is implemented by using good fitting PBFM. The proposed approach can quickly track any given reference signal without the need of a mathematic model of chaos system. The numerical simulation results show that the proposed control method has not only the fast response speed and high control accuracy, but also a strong inhibitory ability to parameter perturbation and the anti-interference ability for the chaos system.