To solve the problem of large calculation errors in deep penetration shielding calculation, increase its computational efficiency and reduce computational variance. This study is based on the CADIS theory and uses Monte Carlo adjoint calculation to process the adjoint calculation results into weight window parameters and source bias parameters within each grid in space, which are used to guide the Monte Carlo method in reducing variance calculation and achieve a new method of self coupling shielding calculation for the Monte Carlo method. This study demonstrates through benchmark problem calculation cases that the Monte Carlo self coupling method based on CADIS theory not only eliminates the complex processing of coupling deterministic and Monte Carlo methods, but also has high computational efficiency and accuracy. Therefore, this study establishes a Monte Carlo self coupling variance reduction method based on CADIS theory, which can be used for accurate calculation of deep penetration shielding problems.