Abstract: BZ oilfield is located on the south fault-terrace belts of Shijiutuo rise,its accumulation condition is superior and Dongying formation of Paleogene has developed a large number of high quality reservoirs.But due to the sand-shale interbed structure and insufficient seismic data resolution,the reservoir identification is difficult to be realized.In combination with seismic attribute and spectral decomposition technique,the reservoir quantitative prediction has been realized.First,it starts to analyze the correlation of reservoir lithological parameters and attributes based on the effectiveness of seismic attributes to reservoir identification,and by using the method of attribute optimization and cluster fusion to predict the distribution of reservoir.Meanwhile,by using spectrum decomposition technology to compute the first peak frequency of reservoir and building the velocity model under the constraint of structure,then predicting the thickness of reservoir by the relationship between the thickness of the strata and the tuning frequency and the formation velocity,and finally correcting its thickness with the thickness from drilled wells.With this method,good results have been obtained in BZ oilfield reservoir prediction.