Abstract: Forward modeling indicates that the magnetic anomalies in deep volcanic basin are usually not big, and occur as secondary anomalies superimposed on a strong background. Based on the integral iteration continuation method, we proposed in this paper a new method, i.e. the integral iteration continuation level curve, for identification of deep basin volcanic rocks from the magnetic anomaly information. This method can not only enhance the magnetic anomalies in the deep volcanic basin, but also eliminate the effects of shape and size of volcanic rocks in different depth to achieve balanced magnetic anomalies. Application of this method is successful to the deep volcanic rocks in the Songliao Basin.