Abstract: The reservoir in well J of Hangjinqi block is thin and characterized by dense lithology and strong heterogeneity. The resolution of seismic data is low and it is difficult to separate the top and bottom interfaces of sandstone. Meanwhile, due to the strong reflection of coal measure strata and the small difference of wave impedance between sand and mudstone, the seismic response characteristics of sand body are not obvious and the reservoir prediction is extremely difficult. Therefore, a set of reservoir prediction process for coal measure strata in thin sandstone of this area is proposed. Firstly, through the comprehensive analysis of drilling, logging and seismic data, the sensitive parameters of interpretation lithology were determined and the lithology interpretation template was established. Secondly, the geostatistical inversion method based on Markov chainMonte Carlo algorithm was used to locate the thickness and spatial distribution of coal seam from strong reflection, and then the lithology data body was obtained through the neutron attribute co-simulation of cloud transformation. Finally, combined with the lithology interpretation template, the effective prediction of thin reservoirs in coal measures was realized.