Abstract: The fluvial sandy reservoirs are usually quite thin in thickness and changing rapidly in both vertical and lateral directions in the Qinhuangdao Oilfield. It is difficult to identify those reservoirs with ordinary seismic inversion methods due to limited frequency bands. The geostatistical inversion method based on MCMC algorithm, however, is more powerful in improving resolution and accuracy for reservoir prediction, and the application of facies controlling method during the process may enhance the facies model constraints on reservoir identification to help developing a quantitative model from a conceptual model, and establishing the reasonable 3D porosity, permeability and oil saturation models. The recommended workflow is, firstly, to reconstruct a pseudo p-wave impedance curve using the lower and higher frequency components of SP and deep resistivity logging for distinguishing the sandy and shaly layers more sensitively; then, to take the pseudo p-wave impedance as intermediate variable and analyze the spatial variation of sand-shale lithology and physical properties using the double-variogram function; and finally, to make geostatistical inversion based on lithological facies controlling MCMC algorithm. The 3D porosity model achieved with vertical resolution of 1-ms can not only reveal the superimposed relations of the sand-bodies in different stages, differentiate the distribution of intercalations in vertical sequences and define the spatial pattern of sedimentary cycles more precisely, but also can make it clear the variation in physical properties of the meandering channel sands, and reduce the multiple-solutions of seismic data effectively.