Abstract: A new wavelet ridge and instantaneous frequency extraction method based on maximum gradient is presented. Firstly, continuous wavelet transform is performed on non-stationary response signals, and as a result, the vectors time, scale and wavelet coefficients modulus constitute a three dimensional coordinate system (x, y, z). Secondly, the searching step is conducted from the initial point of the wavelet ridge, and meanwhile the distance in the time-scale plane (x-y) and height direction (z) and the gradient of wavelet ridge are calculated. Finally, the wavelet ridge curves can be successfully extracted by connecting the points with the maximum gradient. The effectiveness and accuracy of the proposed method are validated via numerical examples and a time-varying cable test. The results demonstrate that the method can extract the instantaneous frequency effectively and that its identification accuracy is superior to the method based on the maximum of wavelet transform modulus and dynamic optimization.