There are two methods for identifying formation interface in oil wells: core analysis, which is a precise approach but costly and time consuming, and well logs analysis, which petrophysists perform, which is subjective and not completely reliable. In this paper, a novel coupled method was proposed to detect the formation interfaces using GR logs. Second approximation level (a2) of GR log gained from optimum mother wavelet decomposition was used for formation interface detection. Short time Fourier transform (STFT) of a2 was gained since the window band was fixed in the entire of well depths. Inverse STFT of various windows of transformed data was gained, which creates various signals in depth domain. To this end, a novel formulation was developed to obtain modified signal for formation interface detection. The mean of various resulted signals creates a smooth signal the logarithm well of which highlights formation interfaces. Synthetic data were used to test the applicability of proposed algorithm. Accordingly, GR logs corresponding to five different wells located in an oilﬁeld in south of Iran also were used to investigate the accuracy and applicability of the proposed method. Lastly, the validation process took place by comparing the results of core data analysis and the proposed method. Good agreements were obtained between these approaches, demonstrating the applicability of the proposed methodology.