Abstract: This paper presents GR curves inversion for thin reservoir characterization in the studied area, using the series inversion of multi? attribute regression (MAR) and probabilistic neural network (PNN) based on the geophysical response analysis of reservoir. The result shows that the sand body prediction accords with the whole sedimentary features in the studied area, with high vertical resolution, clear boundary of lateral sand bodies. It could properly reflect the distribution of reservoirs and can be as a guide for next petroleum exploration in this area