Abstract: Fractured reservoir is of vital importance in oil and gas exploration However, because of their non-isotropic characters and 1ateral variations, the identification and prediction of fractured reservoirs are often difficult. Currently, most seismic approaches are still in a trial stage. In this paper, we will describe two seismic approaches that are applicable in solving the problem, i.e. coherence analysis and synthetic runic log technologies. The essential of coherence analysis is to as the corrdation or similarity between seismic traces. The distribution pattern of fractured reservoirs can be identified after the processing of 3D seismic data by coherence analysis. Based oil the sensitivity of the studied fractured reservoir, we further calculated, eight seismic parameters from 3D coherence date volumes and selected the minimum, seismic parameter from these parameters. The purpose of synthetic sonic logging is to convert seismic datas into bedded log sections by extracting the interval velocity information. Three critical technologies have been applied in this process based on synthetic seismogram’s scaling. They are frequency-domain deconvolution, phase correction and low-frequency modeling. The interval velocities that are consistent with the reservoir development section are extracted after the synthetic sonic data volume is obtained. These data are then presented as plane maps. Their trends reflect the distribution of fractured reservoirs. These two seismic approaches are applied to 3D seismic data volumes in the predication of fractured reservoirs that are developed in the Paleozoic carbonates and the Archaean granite-gneiss in southern Bohai basin. The predictions have been further tested by drilling exploration.