Abstract: For the nonhomogeneous condition of space-time adaptive processing(STAP),a new unscented space-time adaptive processing algorithm is proposed that uses the unscented transformation(UT) to obtain the approximative covariance matrix in the nonhomogeneous condition,namely,the transformation from the original nonhomogeneous data to its corresponding homogenous data is thought as the special case,and by the proposed algorithm,the covariance matrix of the homogeneous data that transformed from the original nonhomogenous data can be estimated,the impact of the nonhomogeneous characteristic is reduced effectively.Simulation results indicate that the proposed algorithm has the favourable performance for moving targets detection.