The invention relates to the field of radar signal processing, in particular to a dynamic channelized subband frequency spectrum detection method based on eigenvalues. The method comprises the steps of according to an ith subband signal of a channelized output, through multi-channel conversion of a single-channel signal, obtaining an MXN-dimensional observation matrix, and constructing a samplingcovariance matrix; carrying out feature decomposition according to the sampling covariance matrix to obtain an average eigenvalue in a corresponding form and a current subband minimum eigenvalue, andconstructing a detection statistics amount of a corresponding algorithm; obtaining an expression of a detection threshold of the corresponding algorithm through a false alarm probability according tothe actual condition; and determining whether the signal exists or not according to the judgment expression of the corresponding detection algorithm, namely, when alpha is larger than gamma, judging that the signal exists, otherwise, judging that the signal does not exist. Compared with an existing classic frequency spectrum detection processing method, under the conditions of low signal-to-noiseratio and few sampling points, the higher detection performance is obtained, the accuracy of detection is improved, and the requirements of a signal electromagnetic environment in a future electroniccombat are better met.