Abstract: Fractures serve as important seepage channels and storage spaces for hydrocarbon reservoirs. Therefore, the precise prediction of their distribution characteristics (density and orientation) holds critical significance for hydrocarbon exploration and development. However, existing fracture prediction methods using post-stack seismic data (such as coherence and curvature) cannot effectively extract azimuthal differences, thus failing to meet the demand for high-precision fracture prediction. Given this, this paper proposed a fracture prediction method by performing ellipse fitting on azimuthal attribute differences of post-stack seismic data. Specifically, the original data were optimized using structure-oriented filtering, from which three or more characteristic azimuth angles were further selected to extract azimuthal attributes. Then, the extracted attributes were analyzed using ellipse fitting to quantify attribute anisotropy. Ultimately, the fracture density and orientation were inverted using ellipse parameters (the ellipticity and major/minor axis directions). The practical application of this method in an oil field demonstrates a high coincidence between the predicted fracture density and direction and the actual drilling data, achieving superior accuracy compared to conventional fracture prediction methods. This method provides a high-precision fracture prediction solution for areas lacking wide-azimuth pre-stack seismic data.