A problem of structure identification of continuous nonlinear dynamical systems on the set of continuous block-oriented models is considered. Methods of structure identification in steady state based on the observation of the system’s input and output variables at the input sinusoidal influences, at the input periodical influences, having Fourier’s uniformly and absolutely convergent series, and at the input random processes with normal distribution are proposed. The criteria determining the model structure are developed. The structure identification methods are investigated by means of both the theoretical analysis and the computer modelling.