The invention discloses a short ring distribution based weighted message passing decoding method which is performed through the following steps: 1) initializing; 2) counting the number of rings with ring length (g) from each edge in a check matrix (H) wherein the ring length (g) represents the shortest ring length in a Tanner graph for H; 3) calculating the weighted parameter (wnm) for each edge in the check matrix (H) wherein wnm represents the weight of the edge (Enm) between a variable node (vn) and a check node (cm); 4) calculating the value of information transmitted from the check node to the variable node; 5) calculating the weighted value of information transmitted from the variable node to the check node; 6) calculating the correction factor vector represented in the figure; 7) if the vector is zero or if the number of current cycles is more than the allowed number, then going to step 8; otherwise, switching to step 4 and with one more cycle time; and 8) outputting codes. The method of the invention fully considers the structural differences inside an LDPC code, raising the decoding performance.