Abstract: Wavelet analysis is a rapidly developing domain of applied mathematics. It has been used in turbulence, and several methods are presented to study the coherent structure in turbulence. In this paper, two conclusions are made from the relation between wavelet energy spectrum and Fourier energy spectrum. The first conclusion is that the peak height of wavelet energy spectra of sine and cosine waves is proportional to their scale, and the second is that the peak width of wavelet energy spectra of sine and cosine waves is proportional to their scale. Based on wavelet analysis and these two conclusions, a new energy maxima criterion of coherent structure, P(a m)/ a m=max a >0 P(a )/ a , where P is wavelet energy spectrum, which can extract several coherent structures of close scale in turbulence and determine their scales, is presented. At the same time, Morlet wavelet is found to have a higher resolution ratio than Mexican hat wavelet and some orthonormal wavelets, such as Daubechies, Symlets, Meyer and Coiflets, which can distinctly distinguish fundamental wave from subharmonic. So Morlet wavelet is much suited to detect coherent structure in turbulence. The new criterion is validated with simulative signals which are constructed by adding three kinds of noises with different energy spectra exponents (r=0, 1.00, 1.67) to y=sin(16π x )+sin(32π x ) and keeping the intensity of coherent structure about 0.1. The results show that the new criterion can distinguish fundamental wave from subharmonic based on simulative signals with a small coherent structure intensity. At last, the new criterion is validated with three-dimensional fluctuation velocities of a round jet flow near the nozzle exit. The results show that the scale of coherent structure determined by the new criterion is equivalent to the integral scale, and the coherent structure is a large vortex.