While the polynomial fuzzy model can be viewed as a more effective model with accuracy to describe a nonlinear system to be controlled, unlike the widely used T-S fuzzy model, the so-called separation property does not hold anymore when it comes to that the control state is unavailable. It is more challenging to design a state observer, let alone a disturbance observer. In this paper, after observers of a so-called lumped disturbance and the control state are proposed, a fuzzy controller is designed employing the observers in order to improve the control performance. Also, a computer simulation example is provided to illustrate the effectiveness of the proposed approach.