Stratigraphic sequence interpretation and correlation are part of basic geologic research, but present frequent problems such as subjective and accurate division and correlation of sequence cycles, and a multiplicity of solutions to high-frequency sequences. We developed a novel method, termed frequency trend attribute analysis (FTAA), to solve these problems and improve the accuracy of division. The method was based on maximum entropy spectrum analysis data, built on theoretical foundations, and tested on geologic models as well as empirical data. We developed examples of how FTAA can improve stratigraphic division and correlation. We extracted frequency trend lines from well logging data (using all or a selected part of a facies-sensitive log such as the natural gamma-ray log) whereby the FTAA outcome reflected the overlay series and cycle structures. The resulting frequency trend lines also indirectly reflected changes to the sedimentary environment and base level, and the precise stratigraphic division and isochronous comparisons were automatically deduced from the frequency trend lines. According to the practical comparison with wells in the field, the frequency trend lines were found to be more accurate than using outcrop data, and the method proved to be effective and convenient in use. The FTAA significantly improved the precision and accuracy of automatic division and correlation of sequence cycles.