Abstract
The analytical description of regular LDPC (Low-Density Parity Check) codes correcting ability has been investigated. The statistical dependencies for the maximum number of corrected bits per the code word as a function of LDPC code word length and code rate are given based on multiple experimental analyses of LDPC check matrices. The analytical expressions are proposed for the cases of linear, exponential and polynomial approximations of given results. The most exact analytical formula is proved by criterion of the minimum divergence between the experimental and theoretical results.