In view of the problems of strong longitudinal heterogeneity, poor pore structure, complex porosity-permeability relationship, and that permeability calculated by conventional porosity-permeability can't meet the requirements of fine productivity prediction for low-porosity and low-permeability reservoirs of medium-deep formation in the sea area, the reservoir are divided into several micro-scale lithofacies units based on the identification of lithology and sedimentary bedding structure by logging data of high-resolution electrical imaging combined with diagenesis, which effectively depict the characteristics of longitudinal heterogeneity and relatively uniform and good porosity-permeability relationship of the reservoir. Under the constraints of the lithofacies unit framework, the reservoirs are divided into 6 types of flow units with the reservoir quality factor and lithofacies (equivalent to flow unit) index. Using logging interpretational results of physical property and core permeability analysis data to obtain static permeability of various reservoirs, the accuracy has increased by 44，. On this basis, the conversion model of static permeability and dynamic permeability is established, and the dynamic permeability is applied to the method of the steady-state flow productivity prediction. The fine productivity prediction of Paleogene Huagang Formation reservoir of N structure in Xihu Sag, East China Sea Shelf Basin has achieved good application effect, and the coincidence rate is more than 85.9， compared with DST test results.