Abstract:Having analysed the fuzzy relations between amplitude,phase,frequency,structure curvature,interval velocity,apparent polarity or the thickness of low velocity layer and hydrocarbon accumulation,the authors established the membership functions of these kinds of seismic information.using different decisions and ,1 horizontal cluster sorting,high,low and non-productive areas can be respectively predicted,thus bringing reliable data for locating well sites.The seismic data at 10 wells on a structure in Sichaan Province were taken as an example for experimen t.The decision and sorting were made according to above method,and the seismic data at other 7 wells were used to check the experiment.The high,low and non-productive areas in a part of the structure were quite successfully ascertained.It is shown that this method can also be used to predict hydrocarbon accumulation area.The superiorides of this method to others(for example,mathematical statistics method) are as follows,(1)it not only predicts productive area but also separates high-productive area from low-productive area,and (2)it favours achieving automatic hydrocarbon indication.