Flutter solution is an eigenvalue problem which has variable parameters (such as air speed). In flutter mode identification and flutter prevention, the correspondence between the baseline and perturbed eigenvalues due to parameter variation should be determined accurately, and mode crossing should be avoided. For mode tracking in flutter solution ��p-k�� procedure, the forecast tracking method was developed, which was based on the perturbation theory of eigenvalues. The method used the baseline eigenvalues and the left and right eigenvectors to estimate perturbed eigenvalues, and sorted modes by their estimated values. The forecast tracking method was compared with two existing methods, one was similarity sorting method, the other was orthogonality check method. Two testing cases were used to validate the effectiveness of the methods. The numerical results show that the forecast tracking method is better than the existing methods in effects of mode tracking.