The control algorithm for nonlinear time-invariant systems with compensation of unknown disturbances under measurement noises is proposed. The dimension of noises is equal to the state vector dimension of the plant. Disturbances can be presented in any equation of the plant model. The novel control law is based on the noise estimator and the disturbance compensator. The accuracy in the steady state depends on the first derivative of disturbance and the smallest component of noise vector. A sufficient condition in terms of linear matrix inequality provides feasibility of the proposed algorithm. The simulations show the efficiency of the proposed method compared with some existing ones.