Abstract: Low-porosity low-permeability reservoirs are widely developed in the Paleogene Q structure in the East China Sea Basin, and the overall heterogeneity of sand body curtain changes dramatically. The key problem to be solved in the integration research of exploration and development of the Q structure is to delineate the superior reservoir development area of "sweet spot". A prediction method of seismic "sweet spot" was established by sensitive attribute inversion, of which the main idea is to realize the prediction of tight reservoir "sweet spot" by advanced optimization in seismic lithofacies, physical facies, and fluid facies. In this paper, the EEI (extended elastic impedance) attributes combining AVO (amplitude variation offset) intercept and gradient were used to predict lithofacies. The low-density characteristics of high-porosity sandstone reservoir were used to predict physical facies. Combining qualitative and quantitative analysis in rock physics, new composite attributes were optimized to identify fluid facies. Finally, "sweet spot" reservoir could be predicted accurately under the guidance of hierarchical phase control. Application examples showed that this method could be effectively characterize the distribution of tight clastic "sweet spots" in the research area, and provided an important reference for subsequent target search and well-location optimization in high reliability and application value.